A Review of Artificial Intelligence Techniques for Low-Carbon Energy Integration and Optimization in Smart Grids and Smart Homes

Abstract

The growing demand for electricity in residential sectors and the global need to decarbonize power systems are accelerating the transformation toward smart and sustainable energy networks. Smart homes and smart grids, integrating renewable generation, energy storage, and intelligent control systems, represent a crucial step toward achieving energy efficiency and carbon neutrality. However, ensuring real-time optimization, interoperability, and sustainability across these distributed energy resources (DERs) remains a key challenge. This paper presents a comprehensive review of artificial intelligence (AI) applications for sustainable energy management and low-carbon technology integration in smart grids and smart homes. The review explores how AI-driven techniques include machine learning, deep learning, and bio-inspired optimization algorithms such as particle swarm optimization (PSO), whale optimization algorithm (WOA), and cuckoo optimization algorithm (COA) enhance forecasting, adaptive scheduling, and real-time energy optimization. These techniques have shown significant potential in improving demand-side management, dynamic load balancing, and renewable energy utilization efficiency. Moreover, AI-based home energy management systems (HEMSs) enable predictive control and seamless coordination between grid operations and distributed generation. This review also discusses current barriers, including data heterogeneity, computational overhead, and the lack of standardized integration frameworks. Future directions highlight the need for lightweight, scalable, and explainable AI models that support decentralized decision-making in cyber-physical energy systems. Overall, this paper emphasizes the transformative role of AI in enabling sustainable, flexible, and intelligent power management across smart residential and grid-level systems, supporting global energy transition goals and contributing to the realization of carbon-neutral communities.

1. Introduction

Global energy demand continues to rise alongside residential, industrial, and economic growth, reinforcing energy consumption as a major source of CO₂ emissions. Evidence from 31 advanced and emerging economies shows that although emissions have partially decoupled from economic growth, particularly in advanced economies, the link between energy consumption and CO₂ emissions has intensified, indicating that current energy use patterns remain carbon-intensive and require cleaner energy production pathways [1]. In particular, building operations account for about 30% of global final energy consumption and 26% of energy-related CO₂ emissions, and despite declining direct emissions in 2022 and expanding efficiency standards, faster and more widespread action is required this decade to align the sector with the Net Zero Emissions by 2050 pathway [2]. As urbanization continues, especially in emerging economies, the demand for energy in residential buildings is expected to increase. In response, Figure 1 illustrates the projected rise in the share of electricity in total final energy consumption under the Net Zero Scenario from 2005 to 2030, highlighting electrification as a key pathway for improving building-sector sustainability [3]. To meet these challenges, the global transition to cleaner energy sources has become a pressing priority worldwide, driven by the urgent need to curb accelerating global warming and avert its increasingly severe and far-reaching consequences [4].

Smart homes have emerged as a promising solution for optimizing energy consumption amid rising global demand. By integrating interconnected devices and sensors, these homes enable real-time adjustments based on user behavior and environmental conditions [5]. However, their ability to significantly reduce energy costs and environmental impact remains constrained by limited integration of clean, low-carbon technologies such as solar and wind energy [6]. Key barriers include high installation costs, the intermittent nature of renewable sources, and the need for advanced energy storage to ensure reliable supply [7]. Artificial intelligence (AI) offers a powerful way to address these limitations by automating and optimizing energy systems within smart homes. AI-driven solutions enhance overall efficiency, improve demand forecasting, enable intelligent load shifting, and facilitate better integration of renewables by predicting generation patterns and managing storage effectively [8]. Nevertheless, achieving substantial reductions in the environmental footprint of residential energy use requires not only AI optimization but also the widespread adoption of low-carbon technologies including solar power, wind energy, heat pumps, and geothermal systems [9].

According to the International Renewable Energy Agency (IRENA) [10], renewable energy could supply 77% of global power by 2050, up from 16% in 2020, but this requires the global adoption of clean energy systems and energy-efficient technologies in the residential sectors. One of the major challenges with renewable energy sources (RESs) is their intermittency. Renewable energy growth will be driven mainly by solar photovoltaics (PV) and wind, with nearly 4600 GW of new capacity added between 2025 and 2030, bringing the total renewable capacity to about 9530 GW by 2030. This expansion is expected to raise the renewable share of global electricity generation from 32% in 2024 to around 43% by 2030, with solar and wind together supplying approximately 27% of total generation [11]. Geothermal energy offers a reliable and continuous power supply, with an average utilization rate exceeding 75% in 2023, compared to less than 30% for wind power and less than 15% for solar PV [12]. Although geothermal energy currently meets less than 1% of global energy demand, advances in drilling technologies, particularly horizontal drilling and hydraulic fracturing adapted from the oil and gas sector, are expanding its potential for wider deployment as a clean and dispatchable energy source [12]. Energy storage solutions are essential to address the intermittency of renewable energy sources, enabling surplus energy to be stored and used during high-demand or low-generation periods, further enhancing the resilience of smart homes [13].

Beyond energy storage, improving the energy efficiency of appliances is equally important in minimizing overall residential energy demand and lowering costs [14]. The integration of intelligent, energy-efficient technologies, such as smart appliances, advanced lighting systems, and high-performance HVAC solutions, is essential for optimizing electricity consumption without compromising operational effectiveness [15]. According to the United Nations Environment Programme (UNEP) and United for Efficiency (U4E), improving the energy efficiency of lighting, appliances, and equipment in developing and emerging economies could save up to USD 130 billion per year by 2040, while also contributing to reduced carbon emissions and climate change mitigation [16,17]. Energy-saving practices in smart homes include smart thermostats that learn household patterns, energy-efficient appliances, and LED lighting systems that consume significantly less power [18].

Since residential buildings account for a significant portion of global emissions, reducing household energy consumption remains a key priority in sustainability efforts [19]. Apart from energy conservation, smart homes enhance resource management, with systems like intelligent water management saving up to 30% on water bills [20], while smart appliances adapt to real-time demand, reducing resource use. The global smart home market was valued at USD 183.69 billion in 2024 and is projected to reach USD 949.92 billion by 2032 at a 22.80% CAGR, driven by IoT and AI-enabled automation and energy efficiency demand, with North America holding 36.15% of the market in 2025, Asia–Pacific 29.48%, and smart kitchens leading with a 38.15% segment share [21]. With their ability to reduce utility bills and improve energy efficiency, smart thermostats and lighting alone can reduce heating and cooling costs by 10–15% and electricity use by 20%, respectively [22].

The integration of AI with smart homes helps automate household tasks and predict and manage energy consumption, ensuring that both renewable energy and traditional resources are used efficiently and sustainably [23]. Smart home systems use machine learning (ML) algorithms to analyze historical energy consumption patterns and predict future energy needs. By learning from user behavior and environmental factors, these systems automatically adjust appliance schedules and optimize energy usage in real time, ensuring greater efficiency and cost savings [24]. For example, voice-activated devices such as Amazon Alexa, Xiaomi Mi Home, and Google Home use AI to personalize recommendations, integrating IoT technologies to coordinate energy management across smart devices [25,26,27]. These devices, when integrated with Internet of Things (IoT) technologies, work together to optimize energy management within smart homes. By leveraging IoT, these systems can coordinate the operation of various smart devices, such as thermostats, lighting, and appliances, to manage energy use more effectively [28]. Additionally, they facilitate the seamless integration of renewable energy sources, ensuring that energy consumption is both efficient and sustainable [29]. Furthermore, AI-driven peer-to-peer energy trading platforms enable homeowners to sell excess renewable energy back to the grid, promoting a decentralized, sustainable energy system [30]. AI-driven demand-side management (DSM) strategies help to reduce electricity costs by optimizing energy usage during peak demand periods [29].

Predicting energy consumption is essential for optimizing energy production and distribution in IoT-based smart homes and smart grid applications, but the diversity of household consumption patterns makes accurate forecasting challenging. Moreover, many existing approaches rely on one-step forecasting with limited prediction horizons, motivating this study to evaluate baseline, ARIMA, SARIMAX, and univariate and multivariate LSTM models to improve prediction accuracy [31]. Similarly, a fuzzy logic-based energy management system demonstrates flexibility in handling uncertainties in household demand, EV behavior, and photovoltaic generation, achieving fast real-time performance (52 ms) and successful testing on large-scale distribution feeders, although the complexity of fuzzy rule design and tuning may pose scalability challenges that require expert knowledge in broader deployments [32]. While much progress has been made in integrating clean and low-carbon technologies into smart homes, a gap remains in their optimization through advanced AI models [33]. Specifically, while most existing studies focus on individual aspects of energy efficiency or renewable energy integration, there is a gap in research addressing the collaborative application of AI, IoT, and renewable energy solutions for optimizing energy consumption in smart homes. Without such integration, smart homes may still rely heavily on fossil fuels, limiting their ability to achieve net-zero goals. Additionally, the challenges of scaling these technologies and integrating them into existing infrastructure remain underexplored.

This review addresses key gaps in the optimization of clean and low-carbon energy technologies in smart homes using advanced AI models. Previous studies have mainly examined separate aspects such as energy efficiency, IoT implementation, or individual renewable energy integration. However, this work reviews the literature and proposes a comprehensive framework that combines AI-driven predictive modeling, demand-side optimization, adaptive scheduling, real-time adaptability, and multi-energy renewable systems.

Table 1. Comparative analysis of recent studies on AI-driven energy management in smart homes.

Study	Key Characteristics	Limitations	Our Review Contribution
[34]	Combines AI with IoT for real-time energy optimization in smart homes.	Does not integrate renewable energy sources; scalability not addressed.	Highlights the importance of integrating renewable energy sources for comprehensive energy optimization.
[35]	Systematic review of AI-based energy self-management in smart buildings, organized using the Autonomous Cycles of Data Analysis Tasks (monitoring, analysis, decision-making).	Most existing studies concentrate on decision-making tasks (optimization and control), with limited attention to full autonomous cycles, feature engineering, and multi-agent integration.	Emphasizes the need for AI systems to incorporate renewable energy sources for sustainability.
[36]	Analyzes 93 articles on smart home energy management systems, focusing on architecture, algorithms, and applications.	Many systems are conceptual; limited practical implementation with renewable energy integration.	Suggests pathways for practical implementation and integration of renewable energy in smart home systems.
[37]	Develops a multivariate LSTM model for smart home energy consumption prediction, achieving strong predictive accuracy (MSE of 0.02284, RMSE of 0.15113, MAE of 0.184, MAPE of 0.123, and R2 of 0.694), demonstrating improved performance over previous methods.	Limited to household-level prediction without considering renewable integration, demand response, grid interaction, or carbon impacts; lacks discussion of scalability, interpretability, and real-time deployment.	Highlights the importance of integrating battery storage systems (ESS) for better load management and grid independence, emphasizing real-time adaptability in energy management systems.
[38]	Presents an Optimal Power Management System (OPMS) designed for smart homes in 6G environments. It integrates RESs and battery optimization through scheduling techniques and uses Multi-Access Edge Computing (MEC) to reduce latency and improve the system’s responsiveness.	Faces challenges in large-scale applications, particularly due to the high computational cost and complexity of real-time energy management in extensive smart cities.	Suggests using MEC and AI-based heuristics to improve scalability and efficiency in real-time energy management.
[39]	Systematic literature review from 2018 to 2024 on smart home energy management models. Focuses on energy optimization, AI/ML models, demand-side management, renewable energy integration, and user behavior.	Lack of focus on integrating user behavior and data privacy concerns HEMSs. Does not explore P2P energy trading or local-level energy trading.	Provides a holistic view of smart home energy systems and identifies significant gaps in existing models: lack of user-centric design, real-time adaptability, and privacy concerns. Proposes future directions focusing on AI and ML integration, scalability, and user behavior integration.

shows various approaches to smart home energy management, including the integration of AI with IoT for real-time energy optimization, the use of machine learning (ML) and RES for energy efficiency, and the application of optimization techniques like genetic algorithms and particle swarm optimization. However, several limitations are identified, such as the lack of renewable energy integration, limited practical implementation, scalability issues, and the need for further exploration of real-time adaptability, battery storage systems (ESS), and user behavior integration. This review contributes by emphasizing the importance of incorporating renewable energy sources, enhancing real-time adaptability, and addressing scalability challenges, while also proposing pathways for integrating battery storage and exploring future directions in AI and ML integration. By doing so, it integrates current advancements in AI and energy management, while also highlighting the need for lightweight, scalable AI models to optimize multi-energy systems and support decentralized decision-making in future low-carbon networks.

This study highlights the progress made in integrating low-carbon technologies and the full potential of AI-driven optimization in managing the complex dynamics of energy consumption, renewable energy integration, and real-time adaptability within smart homes. The following are the main contributions of this study:

• Provides a comprehensive analysis of clean and low-carbon energy technologies in smart homes, focusing on energy efficiency, sustainability, and optimization methods.
• Evaluates various optimization techniques, including computational, heuristic, and machine learning-based methods for improving energy management and load balancing.
• Discusses the role of AI and computational techniques in predictive modeling, demand forecasting, and intelligent scheduling for smart home energy optimization.
• Highlights emerging trends such as IoT, blockchain, and decentralized energy systems for smart home energy management.

The rest of this paper is structured as follows: Section 2 presents the methodological approach used in this study. Section 3 provides an overview of clean and low-carbon energy technologies. Section 4 explores the integration of these technologies within smart home environments, highlighting their benefits, challenges, and implementation strategies. Section 5 examines optimization methods used in energy management for smart homes. Section 6 summarizes key findings and insights gained from the literature review, and identifies limitations, providing directions for future research.

2. Materials and Methods

2.1. PRISMA-Based Literature Search and Information Sources

This study followed the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines to ensure thoroughness and transparency, as illustrated in Figure 2. A systematic literature search was conducted across IEEE Xplore, ScienceDirect, PubMed, SpringerLink, Wiley Online Library, Web of Science, and Google Scholar, with the final search completed on 15 August 2025. Database-specific Boolean search strings were constructed using combinations of keywords related to artificial intelligence, low-carbon energy, smart homes, smart grids, and energy optimization, and are fully reported in Table 2. Searches were applied to titles, abstracts, and keywords, and were restricted to peer-reviewed English-language publications published between 2020 and 2025.

2.2. Database Search Strategy

To ensure full reproducibility, database-specific Boolean search strategies were developed and adapted to the syntax of each database. Searches were applied to titles, abstracts, and keywords, where supported by the platform. The core Boolean search string used in this study was as follows:

• “Smart home” OR “home energy management system” OR HEMS OR SHEMS AND
• “Smart grid” OR microgrid OR “demand response” AND
• “Artificial intelligence” OR “machine learning” OR “deep learning” OR “reinforcement learning” AND
• forecast OR predict OR schedule OR “real-time optimization” AND
• “low-carbon” OR decarbonize OR “renewable energy” OR “carbon emission”. Minor modifications were applied to accommodate database-specific indexing rules and search interfaces. Due to platform limitations, Google Scholar searches were restricted to title-based queries using simplified keyword combinations. The complete database-specific Boolean search strings applied, search fields, year filters, and final search dates are summarized in Table 2.

Figure 2. Systematic Representation of the PRISMA Model Strategy.

Figure 2. Systematic Representation of the PRISMA Model Strategy.

Table 2. Database search strategy and search dates.

Database	Search Fields	Boolean Search String	Year Range	Final Search Date
IEEE Xplore	Title, Abstract, Keywords	(full Boolean string)	2020–2025	15 Aug 2025
ScienceDirect	Title, Abstract, Keywords	(full Boolean string)	2020–2025	15 Aug 2025
Web of Science	Topic	(full Boolean string)	2020–2025	15 Aug 2025
SpringerLink	Title, Abstract	(full Boolean string)	2020–2025	15 Aug 2025
Wiley	Title, Abstract	(full Boolean string)	2020–2025	15 Aug 2025
PubMed	Title, Abstract	(adapted string)	2020–2025	15 Aug 2025
Google Scholar	Title only	simplified string	2020–2025	15 Aug 2025

Table 2. Database search strategy and search dates.

Database	Search Fields	Boolean Search String	Year Range	Final Search Date
IEEE Xplore	Title, Abstract, Keywords	(full Boolean string)	2020–2025	15 Aug 2025
ScienceDirect	Title, Abstract, Keywords	(full Boolean string)	2020–2025	15 Aug 2025
Web of Science	Topic	(full Boolean string)	2020–2025	15 Aug 2025
SpringerLink	Title, Abstract	(full Boolean string)	2020–2025	15 Aug 2025
Wiley	Title, Abstract	(full Boolean string)	2020–2025	15 Aug 2025
PubMed	Title, Abstract	(adapted string)	2020–2025	15 Aug 2025
Google Scholar	Title only	simplified string	2020–2025	15 Aug 2025

2.3. Study Screening and Selection Process

All retrieved records were exported into a reference management system, and duplicate records were identified and removed prior to screening. The remaining records underwent a two-stage screening process. First, title and abstract screening was conducted to exclude studies that were out of scope, non-peer reviewed, unrelated to AI-driven low-carbon energy management or not focused on smart homes or smart grids. Second, full-text screening was performed to assess eligibility in greater detail. Studies were excluded at this stage due to insufficient methodological detail, lack of relevance to smart home or smart grid applications, absence of AI-based optimization techniques, or failure to address low-carbon or renewable energy integration.

The number of records identified, duplicates removed, studies excluded at each screening stage, and the reasons for full-text exclusion are explicitly reported in the PRISMA 2020 flow diagram in Figure 2.

2.4. Eligibility Criteria

Eligible studies were selected based on predefined inclusion and exclusion criteria relating to publication period, study type, language, research focus, and technological relevance. These criteria ensured that only high-quality and relevant studies addressing AI-driven optimization of low-carbon energy technologies in smart homes were included. The detailed eligibility criteria are summarized in

Table 3. Criteria for inclusion and exclusion of studies in the review.

Criteria	Inclusion	Exclusion
Publication Date	Studies published between 2020 and 2025	Studies published before 2020 or after 2025
Study Type	Peer-reviewed journal articles, conference papers, and systematic reviews	Undergraduate projects, master’s theses, PhD dissertations, and non-peer-reviewed sources
Focus Area	Studies focusing on AI models for optimizing energy management in smart homes	Studies unrelated to energy management or smart homes
Language	Published in English	Non-English publications
Relevance to Review	Studies addressing AI in the context of low-carbon energy technologies for smart homes	Studies focusing on AI applications outside smart home energy management or clean energy
Technology Focus	Studies addressing AI models (e.g., ML, DL, and PSO) integrated with renewable energy sources	Studies that do not address AI, renewable energy, or smart home systems
Geographical Scope	Global studies with relevance to urban or residential smart homes	Studies focused on non-residential buildings or geographic regions not applicable to smart homes
Energy Optimization Focus	Studies analyzing energy efficiency, renewable energy integration, and sustainability	Studies not related to energy efficiency or low-carbon technologies

.

2.5. Study Quality and Risk-of-Bias Assessment

To enhance the credibility of the evidence synthesis, a study quality and risk-of-bias assessment was conducted prior to final inclusion. Each eligible study was evaluated using a structured assessment framework focusing on key methodological aspects, including clarity of objectives, adequacy of data sources, transparency of AI model design, validation strategy, and reproducibility of results. Studies demonstrating insufficient methodological transparency, weak validation procedures, or unclear experimental design were excluded during full-text assessment. This quality screening ensured that the conclusions drawn in this review are based on robust and reliable evidence.

3. Clean and Low-Carbon Energy Technologies

Smart homes can leverage a variety of clean and low-carbon energy technologies to enhance energy efficiency and reduce dependence on fossil fuels [40]. These include various renewable energy systems, energy storage solutions, and energy efficiency strategies, all of which play critical roles in transitioning away from fossil fuels toward a sustainable energy system, which is thoroughly discussed afterwards.

3.1. Renewable Energy Technologies

Renewable energy technologies refer to the various systems and processes that harness energy from renewable sources, such as solar, wind, geothermal, and biomass, to produce electricity, heat, or fuel. These technologies include solar PV systems, wind turbines, geothermal heat pumps, and bioenergy systems, each of which plays a pivotal role in reducing reliance on fossil fuels and supporting the transition to a sustainable, low-carbon energy system. For instance, solar and wind can provide power generation, while geothermal and biomass can handle heating and cooling. This ensures that each technology contributes uniquely to the overall smart home system.

Renewable energy policies, such as government incentives, subsidies, and regulations, play a crucial role in shaping the adoption and integration of sustainable energy solutions, including renewable energy technologies, in smart homes. Energy policy uncertainty, as noted by Korkut Pata [41], can significantly influence investment decisions and the widespread adoption of renewable energy systems, highlighting the importance of stable policy environments for encouraging homeowners to adopt renewable systems, including solar PV panels and wind turbines. Tsimisaraka et al. [42] applied the CS-ARDL econometric method to data from the top ten OBOR CO₂-emitting countries between 2004 and 2019 to examine how financial inclusion, ICT, renewable energy, globalization, and economic growth affect emissions, finding that renewable energy and ICT reduce CO₂ emissions in both the short and long term, while economic growth increases emissions and globalization lowers emissions only in the long run. Complimenting these findings, Reinoso-Avecillas et al. [43] emphasized the importance of financial incentives in promoting renewable energy technologies. Their study highlighted how well-designed incentive schemes can facilitate the adoption of non-conventional energy sources such as solar and wind in residential settings. This is particularly significant for smart homes, where energy management systems (EMSs) optimize the utilization of generated renewable energy.

As the global push for sustainability intensifies, integrating clean and low-carbon energy technologies such as PV systems, wind energy solutions, energy storage systems, and high-efficiency heating and cooling systems into residential environments, specifically smart homes, has emerged as a critical solution for reducing energy consumption and mitigating climate change [18]. Recent comprehensive reviews by Bamisile et al. [44] have highlighted that while prior studies often examined hybrid renewable energy systems (HRESs) or energy storage systems (ESSs) separately, few have addressed their integrated optimization, underscoring the effectiveness of hybrid techniques incorporating capacity and CO₂ emissions constraints for sustainable HRES-ESS deployment in contexts like smart homes. Smart homes contribute to smart city energy ecosystems by employing intelligent energy management systems that integrate distributed energy resources (DERs) through advanced technologies such as IoT, machine learning, and data analytics to improve efficiency, reduce costs, and enhance sustainability [45]. Despite the promising potential of clean and low-carbon energy technologies, several challenges persist, particularly in terms of their integration, scalability, and cost-effectiveness. To address these issues, this subsection explores the current state of these technologies in smart homes, identifies the gaps in their integration, and proposes potential solutions based on the latest literature. Among the key technologies, solar, wind, geothermal, and biomass energy systems offer significant potential, each with unique benefits and challenges, which are summarized in

Table A1. Summary of clean and low-carbon energy technologies.

Technology Type	Description	Benefits	Challenges	Gaps	Potential Solutions	Source
Solar Energy	PV panels convert sunlight directly into electricity. Solar thermal systems capture heat for water or space heating. Solar energy is widely regarded as a transformative solution for residential energy needs, reducing dependency on grid power and aligning with sustainability goals. Solar PV cost is approximately USD 0.89–1.01 million/MW.	Renewable and abundant Low maintenance Reduce electricity bills Sustainable and eco-friendly	Intermittent energy source (dependent on sunlight) High initial installation cost Space requirements	Limited capacity for areas with low sunlight or inconsistent weather patterns.	Integration with energy storage systems (ESSs) like lithium-ion batteries or thermal storage to store excess energy for low sunlight periods. Explore BIPV (building-integrated photovoltaics) to address space limitations in urban settings.	[255,256,257]
Wind Energy	Small-scale wind turbines generate electricity by converting the kinetic energy of the wind, often integrated with smart home systems. Wind energy reduces greenhouse gas emissions and enhances energy security. Incentive policies have encouraged adoption despite challenges. Wind (onshore) cost is approximately USD 1.3–2.2 million/MW.	Renewable and clean Low operating costs It can be integrated with storage systems for reliability	Intermittent energy source (dependent on wind) Noise concerns Requires large open space	Inconsistent wind speeds, space constraints in urban areas.	Use Vertical-Axis Wind Turbines (VAWTs) for urban areas. Combine wind and solar energy with ESS to reduce intermittency. Develop predictive algorithms for better integration.	[258,259,260]
Geothermal Energy	Geothermal heat pumps (GHPs) use the Earth’s internal heat for space heating and cooling. Geothermal energy provides a stable energy source, reduces residential carbon footprints, and has a long operational life with low emissions. Geothermal cost is approximately USD 20,000 to USD 30,000 for home systems, and USD 2500 to USD 5000 per kW utility scale.	Reliable and continuous Efficient for heating and cooling Low operational costs	High installation cost Suitable for specific geographical locations Limited scalability in some areas	Not effective in regions lacking natural geothermal resources.	Develop modular geothermal systems with advanced drilling technologies to reduce costs. Combine with solar thermal systems for hybrid solutions.	[258,261,262]
Biomass Energy	Derived from organic materials such as plant and animal byproducts or organic waste. Biomass systems, such as pellet stoves or boilers, offer a sustainable alternative to fossil fuels, particularly for rural areas. Biomass cost is approximately USD 89.21 per MWh.	Sustainable, reduces carbon emissions, supports rural economies.	Resource availability, emissions, efficiency concerns.	Emissions from combustion can affect air quality.	Use advanced biomass conversion technologies like gasification and pyrolysis for cleaner energy production. Integrate IoT-based systems for optimized combustion and minimal waste.	[263,264,265].

of Appendix A.

3.1.1. Solar Energy Technologies

Solar energy technologies, encompassing solar PV systems, when integrated into smart homes, provide clean, reliable energy that directly supports sustainable development by reducing fossil fuel dependence, lowering greenhouse gas emissions, and enhancing environmental protection. Their widespread adoption also drives economic benefits through job creation in installation, maintenance, and manufacturing, while improving energy security and contributing to long-term cost savings for households [46]. However, despite the advantages, solar energy integration in smart homes faces notable challenges in the form of intermittency since the sun does not shine consistently throughout the day or year, leading to periods of low or no energy generation. This makes it difficult to rely solely on solar energy, especially in regions with cloud cover or during winter months [47]. Moreover, the high initial cost of installing solar panels and associated equipment, such as inverters and batteries, remains a barrier to widespread adoption, particularly in urban areas with limited roof space [48]. In the study by Bamisile et al. [49], the efficiency of solar power is significantly influenced by factors such as solar irradiance, ambient temperature, and atmospheric conditions, with solar panels facing reduced productivity in hotter climates, as module temperature increases, decreasing efficiency by 0.4–0.5% per degree Celsius, while atmospheric conditions like clouds, aerosols, and dust can reduce electricity output by up to 60%, especially in desert regions, and long-term changes in solar irradiance driven by climate change and air pollutants present ongoing challenges for maintaining PV efficiency, emphasizing the importance of optimizing PV systems for diverse climates.

To address these challenges, recent studies have emphasized the importance of energy storage systems (ESSs), such as lithium-ion or solid-state batteries, to store excess energy generated during peak sunlight hours for use during periods of low generation [50]. This integration of storage technology enables homes to maintain a continuous power supply despite the variability of solar energy. To solve the challenge of space limitations in urban environments, building-integrated photovoltaics (BIPV) embed solar panels directly into building materials such as roofs, facades, and windows, thereby integrating solar power generation with the building’s architecture [51]. Moreover, technological advancements in solar cells, such as the development of perovskite solar cells, promise to improve both the efficiency and cost-effectiveness of solar power, addressing the current limitations in efficiency and providing a more viable solution for widespread adoption. As perovskite solar cells can be produced at a lower cost while maintaining high efficiency, they represent a significant opportunity for reducing the overall cost of solar energy systems and increasing their accessibility [52].

3.1.2. Wind Energy

Wind energy, particularly small-scale wind turbines, provides another promising renewable energy source for smart homes [53]. Unlike solar energy, wind can be harnessed at any time of day or night, making it an attractive complement to solar energy systems. Small wind turbines, typically rated between 1 kW and 10 kW, can generate electricity for residential use, with excess power being stored or fed back into the grid [54]. However, like solar energy, wind energy faces significant barriers to adoption. Its variability remains a critical issue, as wind speeds fluctuate throughout the day and across seasons, leading to inconsistent power generation [55]. Furthermore, the space requirements for installing wind turbines, especially in urban settings, are considerable. Most traditional and conventional wind turbines need large and open spaces to operate effectively, which limits their use in densely populated cities [56]. Additionally, the noise pollution produced by wind turbines has raised concerns, especially in residential areas, where the turbines’ noise could disrupt residents’ quality of life. The effectiveness of wind energy generation is highly dependent on location, with regions like coastal areas, mountain passes, and open plains being more suitable for wind farms due to higher and more consistent wind speeds, while seasonal variations and topographical features, such as those in Tehachapi, California, where winds are stronger in the afternoon and from April to October, and in Montana, where winter winds are more intense, can affect the reliability of wind power generation throughout the year; in 2023, 10% (425 billion kilowatt-hours) of the total U.S. electricity generation came from wind, with the top five states (Texas, Iowa, Oklahoma, Kansas, and Illinois) producing about 59% of the total wind electricity generation [57].

Recent innovations in Vertical-Axis Wind Turbines (VAWTs), which are wind turbines that generate power through a rotating vertical axis rather than the traditional horizontal axis, have emerged as a potential solution to these challenges. VAWTs are smaller, quieter, and more suitable for urban environments, as they can operate at lower wind speeds and require less space than traditional horizontal-axis turbines, making them ideal for residential or densely populated areas where space and noise are concerns [58]. These turbines are increasingly being explored for integration into smart homes in urban environments, where conventional turbines would be impractical. Moreover, hybrid energy systems that combine wind and solar power, along with energy storage solutions, can help mitigate the intermittency of both sources. By leveraging both wind and solar energy, smart homes can ensure a more reliable and continuous energy supply, even when one of the energy sources is not generating power [59]. Additionally, the integration of advanced predictive algorithms that forecast wind patterns could improve the efficiency of wind energy systems by optimizing the timing of energy storage and consumption [60].

3.1.3. Geothermal Energy: Efficient and Reliable Heating and Cooling

Geothermal energy systems are another crucial component of clean and low-carbon energy solutions for smart homes. These systems use the Earth’s natural heat to provide space heating and cooling through geothermal heat pumps (GHPs). GHPs are highly efficient, with a coefficient of performance (COP) between 3 and 5, meaning that they produce 3 to 5 times more energy than the electricity they consume, making them one of the most efficient technologies for heating and cooling [61]. Despite their high efficiency, the initial installation costs for geothermal systems remain prohibitively high due to the cost of drilling wells and installing heat exchange loops [62]. Furthermore, geothermal systems are highly dependent on geographical location; they are most effective in areas where geothermal resources are readily available, such as regions with natural hot springs or volcanic activity [63]. This geographic limitation makes it challenging to implement geothermal energy systems in areas where such resources are scarce.

Recent developments in modular geothermal systems and advanced drilling technologies offer promising solutions to reduce installation costs. By lowering the cost of installation and making geothermal systems more flexible, these technologies could make geothermal energy more accessible to a broader range of households [64]. Additionally, integrating geothermal systems with solar thermal technologies in hybrid systems can improve their efficiency and reduce overall energy consumption, providing a comprehensive solution for space heating and cooling [65].

3.1.4. Biomass Energy: A Sustainable and Carbon-Neutral Heating Option

Biomass energy, derived from organic materials such as wood, agricultural residues, and food waste, provides a renewable and carbon-neutral alternative to fossil fuels for residential heating. Biomass boilers and pellet stoves are commonly used in smart homes to provide space and water heating. Biomass energy systems have the advantage of being carbon-neutral, as the carbon dioxide released during combustion is offset by the carbon absorbed by plants during their growth [66]. However, biomass energy systems face challenges related to resource availability, efficiency, and emissions. The availability of biomass fuel can be inconsistent, and transportation logistics can add significant costs [67]. Additionally, biomass combustion can release particulate matter and pollutants, which may negatively affect air quality if not properly managed. Advanced biomass conversion technologies, such as gasification and pyrolysis, offer solutions to the challenges of biomass utilization by providing more efficient methods for converting biomass into gas or liquid fuels, which can be used more effectively than solid biomass [68]. The integration of smart sensors and IoT-based control systems in biomass systems can optimize the combustion process, improve energy efficiency, and reduce emissions. These systems can also adjust the combustion parameters based on real-time data from the environment, ensuring optimal performance and minimal waste.

3.2. Energy Storage Systems (ESSs)

Energy storage systems (ESSs) are a key component of residential energy resource management, enabling the cost-optimal integration of photovoltaic generation and electric vehicles in single-family households by storing surplus energy and supporting electricity cost minimization through optimized storage capacity [69]. The recent literature highlights a wide range of energy storage technologies, from chemical systems to mechanical storage solutions, electrical energy storage, and thermal storage technologies, all of which play a critical role in addressing the challenges posed by renewable energy variability. A review study by Li et al. [70] presented the importance of optimal energy storage system (ESS) integration and evaluated the benefits of ESSs within smart grids. They recommended the development of adaptive policies that foster energy storage market growth and improve system integration. They also introduced a software platform that integrated simulations of novel energy generation scenarios with economic evaluations, providing a robust framework for energy storage configuration in smart grids. This approach aligns with the findings of Rahman et al. [71], who explored various energy storage technologies and their applications in electric grids. They reported that these technologies are vital not only for reducing the environmental impacts of fossil fuels but also for enhancing power system reliability and capacity.

Energy storage systems (ESSs) are critical to enhancing the efficiency and reliability of smart homes. Recent developments in ESS technologies, such as advanced batteries and gravity-based systems, provide promising alternatives to traditional methods [13]. These innovations are increasingly being integrated into smart homes, with significant growth in adoption rates. For example, recent studies highlight a rapid increase in the installed capacity of energy storage systems, particularly behind-the-meter (BTM) ESSs deployed at end-user premises, driven by declining costs and the growing integration of renewable energy resources within the transition toward carbon-free smart power systems [72]. The development of ESS technologies is also progressing rapidly. The global market for energy storage solutions has been growing at an annual rate of 15%, driven by advancements in battery storage and hybrid systems [73]. Policies supporting ESSs are vital in this context for which many countries have introduced incentives, including tax rebates, subsidies, and favorable grid connection regulations, to encourage the widespread adoption of ESSs. For instance, in the European Union, storage policy, under the EU Electricity Directive (EU) 2019/944 and REPowerEU amendments, supports the deployment of behind-the-meter energy storage systems, including in residential buildings, through non-discriminatory market access, streamlined permitting, and enhanced grid integration to support the transition to a climate-neutral energy system [74]. Continued research and development in this field will be essential in shaping the future of energy storage and its role in supporting the transition to a low-carbon, sustainable energy system. This subsection explores the key mechanical, chemical, electrical, and thermal energy storage technologies, offering a detailed examination of their current applications, the challenges they face, and the innovative solutions being developed. Drawing from the recent literature, we provide a comprehensive analysis of each technology, highlighting their potential to transform energy management into smart homes and their role in overcoming the barriers to integrating renewable energy sources.

3.2.1. Mechanical Energy Storage

Mechanical energy storage systems, such as pumped hydro energy storage, flywheels, compressed air energy storage (CAES), and liquid air energy storage (LAES), store energy by converting electrical energy into kinetic or potential energy, which can later be reconverted into electricity to support grid balancing and ancillary service [75]. Flywheels use rotational inertia, according to an analysis of ESS studies regarding stabilization (48%) and dynamic storage (52%) [76], while CAES stores energy by compressing air into underground caverns or large containers [77]. These systems are particularly suited for short-term energy storage and grid stabilization, owing to their high round-trip efficiency and rapid response times. In a smart home setting, they can help manage fluctuations in energy demand and provide backup power. However, mechanical storage systems face significant challenges that limit their feasibility for widespread residential use. One of the main issues is the high capital cost and complexity associated with installation. CAES, for example, requires large-scale infrastructure, such as underground caverns, which makes it impractical for use in urban environments [78]. Similarly, while flywheels are more compact, they still face challenges related to cost and space constraints. To address these limitations, research is focusing on developing smaller, more cost-effective CAES systems that could be more suitable for residential applications [79]. Additionally, flywheel technology is evolving, with innovations in lightweight composite materials that could reduce both cost and space requirements. Combining mechanical energy storage with other ESS technologies, such as chemical or thermal storage, could also enhance performance and lower costs, creating hybrid systems that maximize efficiency and reduce dependency on any single technology [80].

3.2.2. Chemical Energy Storage

Chemical energy storage technologies, such as lithium-ion (Li-ion) and sodium–sulfur (NaS) batteries, store energy through electrochemical reactions [81]. These systems are particularly attractive for residential applications due to their scalability and high energy density. Li-ion batteries, for example, are widely used in smart homes, providing a practical solution for storing energy generated by solar panels or wind turbines [82]. They offer the ability to store large amounts of energy in a relatively compact form and discharge it quickly when needed. Despite their advantages, chemical storage systems face several challenges that hinder their widespread adoption. The high initial cost of Li-ion batteries remains a major barrier, and the environmental impact of their manufacturing, particularly due to the use of materials like cobalt and lithium, raises concerns [83]. Moreover, the limited cycle life of batteries means that they degrade over time, leading to increased long-term costs for homeowners. Recent developments are focusing on alternatives such as solid-state batteries, which promise higher energy densities, longer life cycles, and enhanced safety [84]. On the environmental concerns, the use of recycled materials in battery manufacturing is being investigated. Furthermore, emerging technologies such as sodium-ion and zinc–air batteries could provide more affordable, environmentally friendly alternatives to traditional Li-ion batteries [85]. By improving the lifespan, reducing the cost, and promoting sustainable manufacturing practices, these alternatives could make chemical energy storage more viable for smart homes.

3.2.3. Electrical Energy Storage

Electrical energy storage systems, including supercapacitors and superconducting magnetic energy storage (SMES), store energy in electric or magnetic fields [86]. Supercapacitors are particularly effective for managing short-duration, high-power demands, such as those in smart homes during sudden energy fluctuations. SMES, on the other hand, stores energy as magnetic fields in superconducting coils and is highly efficient, offering fast response times that are beneficial for grid stabilization [87]. While electrical energy storage systems provide excellent efficiency, they are constrained by high capital costs and limited power ratings. Supercapacitors, although effective for short bursts of energy, are not suitable for long-term storage needs. Similarly, SMES systems require expensive superconducting materials, making them impractical for residential use. To overcome these barriers, research is focused on improving the energy capacity of supercapacitors and reducing the cost of superconducting materials [88]. The integration of supercapacitors with other storage technologies, such as Li-ion batteries, could provide a hybrid solution that meets both short-term and long-term energy storage needs. By combining the rapid charge/discharge capabilities of supercapacitors with the higher energy densities of batteries, such systems could optimize energy storage in smart homes [89].

3.2.4. Thermal Energy Storage

Thermal energy storage (TES) involves storing energy in the form of heat or cold. Phase change materials (PCMs) and water-based systems are commonly used in smart homes to support HVAC systems [90]. These systems store thermal energy and release it when needed, helping to maintain consistent indoor temperatures while reducing the reliance on conventional heating or cooling methods [91]. PCMs, for example, store and release energy as they undergo phase transitions, such as from solid to liquid. Despite their efficiency, TES systems face challenges, particularly regarding slow response times and limited insulation efficiency. The performance of PCMs is influenced by environmental factors, and water-based systems are dependent on the quality of insulation [92]. Additionally, the energy release and absorption rates of these systems can be too slow for certain smart home applications, such as real-time climate control in living spaces or rapid responses to peak electricity demand from appliances like air conditioning or heating systems, making them less effective in meeting rapid heating or cooling needs. Recent advancements in materials science are addressing these issues by developing high-performance PCMs with faster phase transitions and better insulation properties [93]. Furthermore, M. Khajevand-dalasmi et al. [94] investigated hybrid renewable energy systems by integrating solar thermal technology with wind, photovoltaic, and biomass sources using advanced control and energy management methods, and found that such hybrid configurations improve energy reliability and efficiency while reducing operational costs and greenhouse gas emissions. This combined approach can provide more efficient heating and cooling, and reduce overall energy consumption while improving the comfort and sustainability of smart homes. Examples of different energy storage solutions can be seen in

Table A2. Energy storage systems.

Reference	Type of ESS	Summary	Potential Solutions	Examples
[266,267]	Mechanical Storage	Mechanical energy storage systems, such as flywheels and compressed air energy storage (CAES), store energy in the form of kinetic or potential energy. Flywheels use rotational inertia for short-term energy storage, while CAES use compressed air to store energy in large underground caverns. These systems offer high round-trip efficiency, fast response times, and provide grid stabilization, but are generally constrained by system complexity, size, and higher capital costs.	Develop compact CAES systems, reduce material costs for flywheels.	Pumped Hydro Energy Storage (PHES): Water is pumped to a higher reservoir during off-peak hours and released to generate electricity during peak demand. Compressed Air Energy Storage (CAES): Air is compressed and stored underground, then expanded to generate electricity. Flywheel Energy Storage: Flywheels store kinetic energy in rotating wheels, which can be quickly released as electricity. Mechanical Springs: Springs store energy through compression or tension, often used in smaller applications. Gravity-based Storage: Energy is stored by lifting weights or masses, which can be released to generate electricity.
[268,269,270]	Chemical Storage	Chemical energy storage systems, including batteries (e.g., lithium-ion or sodium–sulfur) and fuel cells, store electrical energy in chemical form and release it through electrochemical reactions. Batteries offer high energy density and fast discharge rates, while fuel cells convert chemical energy directly to electricity. These systems are widely used in smart homes for their high-power density and scalability. However, challenges include higher initial costs, limited cycle life, and environmental impact of materials used.	Improve battery chemistry, reduce environmental impact of materials.	Batteries: Store energy in the form of chemical reactions, such as lithium-ion batteries. Hydrogen Fuel Cells: Store energy in hydrogen, which is converted to electricity through a chemical reaction. Biofuels: Store energy in chemical bonds of organic materials, such as ethanol or biodiesel.
[80,271]	Electrical Energy Storage	Electrical energy storage systems, such as supercapacitors and superconducting magnetic energy storage (SMES), store energy in electric or magnetic fields. Supercapacitors store energy through electrostatic charge, providing rapid charge and discharge cycles, ideal for managing short-duration, high-power demands. SMES utilizes superconducting coils to store energy as magnetic fields, offering high efficiency and fast response times. These systems are effective for stabilizing smart home energy loads but are limited by their high capital cost and power rating constraints.	Develop cost-effective superconducting materials, enhance charge capacity.	Capacitors: Store electrical energy in an electric field. Batteries: Store electrical energy through chemical reactions (e.g., lithium-ion or lead–acid). Supercapacitors: Store electrical energy in an electric field, with high capacitance and rapid charge/discharge capabilities.
[272,273]	Thermal Storage	Thermal energy storage systems store energy in the form of heat or cold. Phase change materials (PCMs) and water tanks are common in smart homes to support HVAC systems, enabling the transfer of thermal energy to meet heating or cooling demands. PCMs store and release energy as they undergo phase transitions (e.g., from solid to liquid), while water-based systems provide seasonal storage capabilities. These systems improve energy efficiency and reduce reliance on grid-based heating or cooling but may have limitations in response time and thermal insulation.	Improve PCMs for quicker thermal response, better insulation for efficiency.	Molten Salt: Stores thermal energy in molten salt, often used in concentrated solar power plants. Ice Storage: Stores thermal energy by freezing water, which can be used for cooling later. Phase Change Materials (PCMs): Store thermal energy through phase transitions (e.g., solid to liquid).

, which highlights the different types of energy storage technologies and summarizes their characteristics.

3.3. Energy Efficiency Technologies in Smart Homes

The integration of energy-efficient technologies in smart homes enhances residential energy efficiency, lowers costs, and reduces environmental impacts. This discussion explores the various advancements and their implications, supported by recent studies. A study by Ukpene and Apaokueze shows that smart home technologies significantly improve residential energy efficiency in Nigerian households, reducing overall energy consumption by 15–25%, with smart appliances achieving the highest efficiency gains of up to 25% [95].

Smart appliances such as refrigerators, washing machines, ovens, air conditioners, and lighting systems are important in optimizing energy usage [37]. Integrated with advanced intelligent algorithms, smart appliances dynamically adjust their operational patterns based on real-time load conditions and user behavior, thereby enhancing energy efficiency. Malysheva et al.’s study demonstrates that energy-efficient technologies achieve measurable efficiency gains, reducing appliance energy use by 10–15%, lowering thermostat setpoints by an average of 1 °C during inactivity, improving Energy Star performance, and decreasing grid dependence through on-site solar generation [96]. These innovations not only lower household energy costs but also contribute to overall energy conservation. In this context, the optimized smart home energy management system discussed by Ali et al. [97] highlights the importance of real-time pricing and hybrid architectures in reducing both grid consumption and costs. By enabling homeowners to adjust their energy usage according to real-time market prices, these systems encourage more efficient energy consumption patterns. Listewnik [98] demonstrates that the combination of energy management systems with renewable energy sources leads to energy savings ranging from 11% to 31%, with some device-specific configurations yielding reductions as high as 60%. This highlights the potential of customized energy management solutions in significantly enhancing energy efficiency in smart homes.

A critical component in this optimization is the IoT, which enables homeowners to remotely monitor and control household appliances. El-Afifi et al. [99] demonstrate that the integration of IoT into Smart Energy Hubs enhances intelligence and efficiency but substantially increases cybersecurity risks, making robust security and resilient data transmission solutions essential before large-scale deployment. Saadawi et al. [100] expand on this by presenting an IoT-based optimal energy management approach that employs Harmony Search optimization techniques to maximize energy efficiency in smart homes, particularly by incorporating renewable energy systems like PV and wind energy.

These systems offer features such as remote control and automated scheduling, which lead to substantial reductions in energy usage. Zhao et al. [101] highlight how smart devices adjust heating and cooling based on user behavior, enhancing both comfort and efficiency. Since solar panels play a key role in reducing reliance on grid power [97], their integration with smart home systems not only promotes energy independence but also aligns energy production with weather predictions, ensuring consistent energy supply [96]. Furthermore, the incorporation of machine learning (ML) algorithms and IoT sensors enables dynamic energy optimization, improving efficiency by adjusting energy use based on real-time data [102]. Energy conservation strategies, such as appliance and user energy profiling and off-peak load scheduling, are also critical components of smart home energy efficiency practices. These strategies have been shown by Fakhar et al. [103] to effectively reduce energy consumption and minimize peak demand, further reducing grid dependency and contributing to both economic savings and environmental sustainability [104]. However, the integration of smart home technologies ranging from energy management systems and IoT-based solutions to smart appliances plays a crucial role in enhancing energy efficiency [105]. These innovations optimize energy consumption, reduce greenhouse gas emissions, and contribute to global sustainability goals. Continued advancements in these technologies, particularly in conjunction with renewable energy systems, will be essential in shaping a low-carbon future for smart home energy use [106]. The evidence suggests that effective deployment of these technologies not only improves energy efficiency but also enhances user satisfaction, offering substantial savings while fostering a more sustainable living environment.

4. Integrating Clean and Low-Carbon Energy Technologies with Smart Homes

The integration of clean and low-carbon energy technologies in smart homes has emerged as a crucial strategy for enhancing energy efficiency, sustainability, and reducing carbon footprints. Recent advancements highlight the potential of these technologies to optimize energy consumption and play a significant role in mitigating climate change. By leveraging RESs such as solar, wind, and geothermal power, smart homes can reduce their reliance on the grid, lower energy costs, and decrease carbon emissions [107]. This integration, when combined with energy storage solutions, ensures energy security by making homes more self-sufficient. Additionally, AI plays a critical role in regulating energy consumption, as advanced algorithms can predict energy demand, adapt to user behaviors, and manage energy storage and distribution in real time, ensuring optimal resource utilization without compromising user comfort [108]. The adoption of smart home technologies has been propelled by several key factors. Technological readiness, as highlighted by Laura et al. [109] is a significant driver, with advancements in AI, ML, and IoT making these systems increasingly reliable and affordable. Moreover, economic incentives, such as government subsidies and rebates, have made the initial investment in smart home technologies more accessible for homeowners, thus accelerating their adoption [110]. Alongside these technological and financial factors, user awareness has played an important role in encouraging the shift toward energy-efficient solutions, with growing concerns about energy consumption and the environmental impact prompting individuals to adopt more sustainable living practices. This heightened awareness is essential for the wider acceptance of low-carbon technologies, as it leads to greater demand for solutions that reduce carbon footprints and promote sustainability.

Despite the significant progress made, several challenges remain. High initial costs and the complexity of integrating multiple technologies, including renewable energy systems and energy management software, continue to hinder broader adoption [111]. Additionally, intermittency issues associated with RESs like solar and wind require effective energy storage solutions to ensure a consistent and reliable energy supply [112]. The challenge of system integration also remains, with the need for seamless communication between AI algorithms, IoT devices, and energy systems. To address these challenges, hybrid energy systems combining multiple RESs and advanced storage solutions offer a promising approach to mitigate intermittency and provide a more reliable energy supply. Furthermore, the development of decentralized energy systems, including P2P energy trading facilitated by blockchain technologies, is emerging as a solution to enhance the efficiency and flexibility of energy distribution, as reported by Gonzalez-Gil et al. [113]. Recent studies by Longwen Chang et al. [114] show that multi-energy complementary systems integrating wind, solar, hydrogen, and battery storage can enhance energy utilization efficiency and system stability under high renewable variability. Although developed mainly for extreme environments, these frameworks provide valuable insights for smart home and smart grid applications, particularly in addressing renewable intermittency, electro-thermal coupling, and coordinated multi-energy operation. They also inform the development of future AI-enabled low-carbon energy management strategies.

In addition to technological advancements, the role of policy frameworks and user engagement cannot be overlooked. Policies that promote energy efficiency, such as tax incentives and rebates for renewable energy installations, are essential to furthering the widespread adoption of smart home technologies. Environmental regulations also drive the transition to low-carbon technologies by incentivizing sustainable practices and reducing barriers to implementation, as reported by Yu and Shi [115]. The successful implementation of smart home technologies requires a holistic approach, combining technological innovation, supportive policies, and active user participation to maximize the environmental and economic benefits of these systems. As the integration of clean and low-carbon energy technologies in smart homes continues to evolve, it is clear that these technologies are integral to achieving sustainability goals [34]. By addressing the current challenges of cost, energy intermittency, and system integration, and fostering greater collaboration between stakeholders, smart homes can contribute significantly to reducing household carbon emissions and optimizing energy consumption. The continued development of scalable, cost-effective solutions, such as smart grids and advanced energy storage systems, will further enhance the role of smart homes in creating a sustainable, low-carbon future.

4.1. Relevance of Clean and Low-Carbon Energy Technologies in Smart Homes

The relevance of clean and low-carbon energy technologies in smart homes has gained increasing recognition as essential for advancing both energy efficiency and sustainability. By incorporating a range of advanced technologies, smart homes are able to optimize energy consumption, reduce carbon footprints, and enhance user comfort [34]. These renewable sources not only contribute to reducing reliance on conventional energy grids but also offer a means of achieving long-term environmental goals. As smart homes evolve, the capacity to seamlessly incorporate renewable energy into energy management systems becomes increasingly vital, ensuring that homes can adapt to changing energy demands while maintaining efficiency and minimizing environmental impact [116].

This aligns with the findings of Tadrak, who discusses the potential of Smart Local Energy Systems (SLESs) to provide cleaner energy solutions by integrating renewable energy generation, management, and usage in residential clusters [117]. Similarly, the development of smart microgrids plays a critical role in modernizing electrical infrastructure and facilitating the integration of RES and energy storage technologies, as emphasized by Patil and Pragati [118]. These microgrids enhance energy efficiency and resilience in smart homes, providing a critical foundation for sustainable energy management. The transition from traditional fossil fuels to RES within smart homes is further emphasized by Chen [119], who highlights the environmental benefits of replacing fossil fuels, which are not only finite but also major contributors to greenhouse gas emissions. This shift is essential for mitigating climate change and fostering a more sustainable living environment.

Furthermore, regarding the adoption of RESs, energy management systems play an indispensable role in optimizing energy use in smart homes. Sanislav introduces a smart platform that monitors and manages energy harvesting from renewable sources, demonstrating the potential for significant reductions in greenhouse gas emissions [120]. Similarly, building on life cycle assessment-based evaluations of smart home technologies, Tippe et al. [121] examine the environmental implications of smart homes, showing that the integration of automation and on-site renewable energy systems can enhance resource efficiency by supporting demand-side management and shaping household energy-use behavior. This notion is further validated by the work of Saadawi et al. [100], who propose an IoT-based approach to energy management that maximizes energy efficiency in smart homes through the optimal utilization of renewable energy systems. Abd-Elsalam et al. [122] propose an intelligent energy management system (EMS) that not only optimizes the use of renewable resources but also manages household appliances to reduce energy costs.

Moreover, the concept of prosumer households that both consume and produce energy has become a key element in the development of smart home systems. Gonzalez-Gil et al. [123] propose a modular architecture for Smart Home Energy Management Systems (SHEMSs) that facilitates interaction between energy consumers and the smart grid. Their work emphasizes the importance of interoperability and security in these systems, which are critical to enabling prosumer-oriented models and fostering a more sustainable energy ecosystem. The convergence of smart technologies, RES, and data-driven solutions not only enhances user experience but also contributes to broader environmental objectives [124]. As research continues to evolve, the potential of smart homes to transform energy consumption patterns and advance sustainable living becomes increasingly clear.

4.2. Implementation of Clean and Low-Carbon Technologies in Smart Homes

This subsection presents the implementation of clean and low-carbon technologies in IoT-enabled smart homes, drawing on system configurations that achieved efficiency improvements of up to 72.3%, energy cost reductions of 61%, and CO₂ emission reductions exceeding 61% [125]. It focuses on the integration of renewable energy systems and low-carbon thermal technologies with automation, connectivity, and real-time energy management. This subsection highlights key technologies such as PV systems, building-integrated photovoltaics (BIPV), energy-efficient appliances, and low-carbon heating and cooling systems as key components of sustainable smart home energy systems.

4.2.1. Photovoltaic (PV) Systems in Smart Homes

PV systems are essential for the seamless integration of renewable energy within smart homes, directly contributing to the optimization of energy usage and the reduction of carbon emissions [126]. These systems harness solar energy, converting it into electricity with high efficiency, thus significantly decreasing dependence on non-renewable energy sources. Recent research emphasizes the growing importance of PV systems not only in enhancing energy independence but also in facilitating grid stability and demand-response capabilities within smart homes [127]. Also, the ability of PV systems to operate in conjunction with advanced energy management frameworks is crucial for optimizing household energy consumption patterns and contributing to the broader adoption of low-carbon technologies in residential energy systems [128]. In this context, grid-tied rooftop solar PV systems with smart metering have emerged as a transformative technology. These systems enable homeowners to sell excess energy back to the grid, creating a more dynamic and sustainable energy exchange [129]. Coupled with advanced energy management systems, PV systems can optimize power distribution between PV generators, electric vehicles (EVs), and the grid, reducing electricity bills and even enabling free EV charging [130]. The integration of rooftop solar PV with EVs offers additional benefits, such as reduced emissions and potential economic advantages through bidirectional charging capabilities, further enhancing the sustainability of smart homes [131]. Moreover, intelligent energy management systems that combine solar energy with battery storage can minimize grid dependency, allowing users to reschedule power flows and sell surplus electricity, thus improving economic efficiency [132]. These innovations highlight the expanding potential of PV systems to create more sustainable, energy-efficient smart homes.

Hua et al. [133] introduced a smart home load scheduling system that utilizes PV generation and demand response strategies to optimize energy consumption, thereby minimizing utility bills and pollutant emissions. Khoiruddin et al. [134] explored the integration of PV technology with smart home systems to enhance energy efficiency and sustainability. It utilizes an ESP32 microcontroller to monitor and regulate power usage based on solar energy availability, allowing users to control appliances via smartphones. The system aims to support environmental conservation by combining PV and IoT technologies in smart homes. Meteier et al. [135] introduced a recommender system for optimizing energy use in smart homes with battery-free solar panels. By predicting energy production and consumption using AI, the system generates recommendations to improve energy efficiency. The study by N. Chinnathambi et al. [136] highlights the challenges in predicting consumption and the potential for IoT-based systems to optimize energy use. Their research presents an energy management system for grid-connected solar PV-powered DC residential buildings. It addresses challenges like energy conversion losses and load dynamics, proposing a system that minimizes these losses and ensures uninterrupted power supply through demand-side management. S. Bahramara [137] proposes a model for managing energy in smart homes equipped with rooftop PVs and energy storage. It addresses the challenge of increased ramp rates in net load due to PV systems and proposes a flexibility-constrained approach to minimize operation costs while maintaining system flexibility. The research by Tutkun et al. [138] aims to reduce electricity costs in smart homes with PV and ESSs. It uses multi-objective optimization to balance cost reduction, user comfort, and peak-to-average ratio, highlighting the impact of dynamic electricity pricing and user preferences on energy management.

4.2.2. Building-Integrated Photovoltaics (BIPV) in Smart Homes

BIPV offer a forward-thinking solution by embedding solar panels directly into building materials like roofs and facades. This technology serves the dual purpose of generating electricity and enhancing both the structural functionality and visual appeal of buildings. Cotfas et al. emphasize the ability of BIPV systems to significantly boost energy efficiency while delivering notable architectural advantages [139]. The integration of BIPV leads to significant reductions in energy consumption for heating and cooling, as these systems may help regulate indoor temperatures while producing renewable energy [133]. In 2023, the global BIPV market was valued at USD 23.67 billion and is projected to grow at a 21.2% CAGR from 2024 to 2030. This growth is fueled by expanding solar PV installations and rising demand for renewable energy solutions [140].

One of the primary advantages of BIPV is their ability to generate electricity while serving as a building component, thus reducing reliance on traditional energy sources. Baghoolizadeh et al. discuss how BIPV systems are likely to be strategically implemented in residential buildings to optimize electricity consumption and production, showcasing their dual functionality as both energy generators and structural elements [141]. This integration is particularly relevant in urban settings, where space is limited, and maximizing energy efficiency is crucial. Jin’s research further supports this notion, demonstrating the potential for BIPV installations on both rooftops and façades of residential blocks, which would be able to significantly enhance energy utilization in urban environments [142]. Moreover, the flexibility of BIPV systems, such as bifacial dye-sensitized solar cells (DSSCs), allows for innovative designs that may adapt to various architectural styles while maintaining high performance under diverse lighting conditions. Barichello emphasizes that these attributes make DSSCs particularly suitable for smart homes, where esthetic considerations are as important as energy efficiency [143]. The ability to customize the appearance of solar cells to blend seamlessly with building designs is a critical factor in promoting the adoption of BIPV technologies.

In addition to esthetic and functional benefits, BIPV systems contribute to the overall energy management strategies of smart homes. Kim et al. highlight the importance of integrating energy management systems that utilize BIPV to optimize energy usage and reduce costs in residential settings [104]. The incorporation of IoT technologies into SHEMSs enhances the efficiency of BIPV applications by enabling real-time monitoring and control of energy consumption [144]. This integration allows homeowners to make informed decisions about energy use, further promoting sustainability. However, the widespread adoption of BIPV technologies faces challenges, particularly regarding high initial costs and the complexity of installation. Chen’s investigation into risk assessment for BIPV projects points out the need for comprehensive strategies to address these barriers, alluding that understanding the risks involved could facilitate better decision-making in residential project development [145]. Additionally, Bošnjaković et al. discuss the importance of stakeholder cooperation and the dissemination of information regarding BIPV technologies to overcome market barriers [146]. BIPV systems offer a promising solution for enhancing energy efficiency and sustainability in smart homes. Their dual functionality as building components and energy generators, coupled with advancements in design and technology, position BIPV as a key player in the transition towards more sustainable urban living.

4.2.3. Low-Carbon Cooling and Heating System in Smart Homes

Low-carbon heating and cooling systems, such as air-source and ground-source heat pumps, utilize renewable electricity or efficient energy extraction methods to provide heating and cooling. Unlike fossil fuel-based methods, which release greenhouse gases when burning fuels like gas or oil, these systems reduce emissions by using electricity from renewable sources or by extracting heat from the environment, thus minimizing their carbon footprint. This system in smart homes marks a substantial leap forward in both energy efficiency and sustainability. These systems utilize cutting-edge technologies such as smart thermostats, advanced heat pump systems (air-source and ground-source), energy-efficient insulation, and automated energy management systems to optimize energy consumption, reduce greenhouse gas emissions, and enhance user comfort [45]. The integration of smart home technologies (SHTs), such as smart thermostats, IoT sensors, and energy management systems, further facilitates the automation of heating and cooling processes [147]. These technologies allow for real-time adjustments based on utility signals, such as changes in energy rates, and individual household preferences, optimizing energy use and enhancing comfort.

Recent studies, such as the one by Ogunjimi et al. [95], highlight the effectiveness of technologies like smart thermostats, advanced heat pumps, energy-efficient insulation, and automated energy management systems in smart homes, particularly in reducing carbon emissions while providing efficient heating and cooling. When paired with RESs, such as PV systems, their potential for enhancing sustainability is even greater. For instance, Larsen et al. [148] demonstrated how SHTs can automate space heating control, enabling flexibility in energy demand and contributing to low-carbon outcomes in district heating networks. For example, Vemuri describes how interconnected devices in smart homes are able to significantly improve energy efficiency by enabling users to monitor and manage their energy consumption effectively. Moreover, Jun et al. [149] explored the implementation of low-carbon smart energy systems in zero-carbon parks, identifying challenges such as energy load management and the instability of RESs. Furthermore, the integration of advanced materials and technologies plays a vital role in enhancing the performance of low-carbon heating and cooling systems. Ohenhen’s comprehensive review of sustainable cooling solutions highlights the importance of innovative materials and smart technologies in improving thermal management and reducing carbon footprints in electronic applications [150]. This emphasis on material efficiency extends to heating systems, where high-performance insulation and energy-efficient components could lead to significant energy savings.

Examples of Low-Carbon Cooling and Heating Systems

Low-carbon heating and cooling systems in smart homes are crucial for sustainable energy use. Examples of low-carbon heating and cooling systems include air-source heat pumps, ground-source heat pumps, water-source heat pumps, electric combi boilers, biomass boilers, micro-CHP systems, and solar water heating, each offering energy-efficient and sustainable alternatives to traditional high-carbon systems like oil- and gas-fueled boilers, with varying costs, efficiencies, and space requirements depending on factors such as climate and building type [151]. The examples of low-carbon heating and cooling systems can be broadly classified into three main categories.

(1)

Heat Pumps in Smart Homes

Heat pumps are a low-carbon and effective method of heating buildings. Rather than burning fuel to generate heat directly, they transfer thermal energy from external sources such as the surrounding air, ground (earth), or water into the home, using electricity mainly to drive the compression process [152]. This fundamental mechanism makes them far more efficient than conventional systems in many scenarios.

The three primary categories of heat pumps commonly applied in smart homes are air-source heat pumps (ASHPs), ground-source heat pumps (GSHPs) (also widely referred to as geothermal heat pumps), and water-source heat pumps (WSHPs), each offering distinct advantages for heating in smart homes, with performance varying based on environmental conditions [153]. ASHPs extract heat from the outside air, making them cost-effective and easy to install, especially in mild climates. However, their efficiency decreases in colder temperatures, which could impact smart home energy management during winter months [154]. In contrast, ground-source heat pumps (GSHPs) leverage the earth’s stable underground temperature, which remains relatively constant year-round (typically 40–70 °F or 4.5–21 °C) regardless of outdoor weather fluctuations [155]. This thermal stability enables GSHPs to achieve higher efficiency than air-source alternatives, particularly in colder climates, while delivering reliable, high-performance heating in winter and cooling in summer [156]. Field data from a GSHP system in Shandong Province showed that optimized operation achieved an average energy-saving rate of 39.66% at load rates above 50%, with additional savings of 7.84% at moderate load conditions (30–50%), highlighting the suitability of GSHPs for smart home systems that leverage intelligent control to reduce operational energy consumption and costs [157]. WSHPs, which extract heat from water sources like lakes or rivers, offer similar benefits to GSHPs but are more site-specific, depending on the proximity to suitable water bodies [158]. They provide higher efficiency in colder climates and can integrate seamlessly into smart home systems designed for energy optimization.

Recent research further highlights the potential of integrated or hybrid configurations to enhance overall performance. For instance Behzadi et al. propose a hybrid system using geothermal energy and a water-to-water heat pump, achieving significant reductions in primary energy use, costs, and CO₂ emissions compared to conventional systems [159]. For example, solar-assisted heat pumps have demonstrated significant enhancements in energy efficiency for residential heating systems. A study by Sezen examined the performance of a solar-assisted air-source heat pump, showing that the integration of solar energy led to substantial reductions in both operational costs and carbon emissions [160]. Similarly, Li et al. explored optimal control strategies for solar air-source heat pump systems, implying their potential to support China’s ambitious carbon neutrality goals by 2060 [161]. These findings reinforce the viability of heat pumps as a cornerstone of low-carbon heating solutions in smart homes.

(2)

Low-Carbon Boilers and Alternatives

Boilers are devices used to heat water or produce steam for heating and hot water in homes. They come in a variety of forms, including the more contemporary, low-carbon combi boiler systems, gas boilers, and the more traditional oil-fired combi boilers [162]. Electric combi boilers and biomass boilers are alternative options, with electric models using electricity to heat water and biomass boilers utilizing organic materials for fuel. Additionally, a micro-CHP (combined heat and power) system can replace a traditional boiler by using gas to simultaneously generate electricity and heat, offering an efficient solution for home energy needs.

Electric combi boilers use electricity to deliver instant space heating and hot water with 99–100% efficiency, producing zero on-site CO₂ emissions and no energy waste. Their compact, tankless design eliminates the need for fuel tanks, making them ideal for smaller homes with limited space. For true zero-emission performance, one should pair them with on-site solar PV panels or choose a renewable-heavy electricity supplier [163]. If there is a need for a very efficient domestic water heating solution, an electric boiler will be a good choice because they do not need to burn fuel, which means that they lose less energy. Biomass boilers, which are powered by burning biomass such as wood pellets from sustainable sources, are an efficient energy solution. If one wants to replace their existing gas or oil boiler with one that works similarly but emits less carbon, a biomass boiler is the best option [164]. Larger homes that require a lot of hot water benefit greatly from these boilers. For homeowners who wish to improve their boiler but do not wish to incur the increased operating costs of an electric combi boiler, they are also a smart option.

Micro-CHP systems, or micro combined heat and power systems, generate heat and electricity simultaneously using the same energy source. Similarly to gas boilers, they are low-carbon and rely on gas mains or LPG. Although they may not provide enough electricity, they reduce energy drawn from the grid [165].

(3)

Solar Water Heating

Solar thermal panels can be used for domestic hot water heating, using solar collectors installed on a house’s roof [166]. In the UK, solar thermal potential is highest in southern regions, especially along the southern coasts of England and Wales, thanks to better annual global horizontal irradiation. Moderately sized domestic systems, with a 4 m² collector area and a 150 L tank, can typically cover up to 70% of annual hot water demand under typical UK conditions, though output varies with season, system size, orientation, and local sunlight [167]. Large-scale solar thermal systems have been successfully deployed under diverse country-specific boundary conditions, supported by appropriate technological solutions and best practice implementations, with Denmark, China, Germany, and Austria collectively accounting for 81% of the total installed capacity worldwide [168].

The future of low-carbon heating and cooling systems in smart homes will likely be shaped by ongoing research and development efforts aimed at optimizing system performance and integrating RESs. For instance, Li et al.’s experimental study on coupled solar and air-source heat pump systems indicates a promising direction for enhancing energy efficiency in rural buildings, signifying that tailored solutions address specific energy needs [169]. Similarly, Khamraev’s work on combined solar heating systems demonstrates the potential for solar energy to meet a substantial portion of residential heating demands, thereby reducing reliance on conventional energy sources [170].

4.3. Energy-Efficient Appliances

Energy-efficient appliances play a vital role in minimizing household energy consumption. Engineered to use less electricity without compromising functionality, these devices offer an effective solution for sustainable living. For instance, Energy Star-certified appliances [171] and building technologies have been shown to reduce household energy consumption and utility costs by approximately 10–12% on average annual bills, with higher savings reported for heating, cooling, and water heating systems, thereby delivering both economic benefits and measurable reductions in residential energy-related emissions [172].

One of the key strategies for enhancing energy efficiency in smart homes is the implementation of HEMSs. These systems are designed to monitor and control household appliances in real time, optimizing their operation based on user preferences and energy consumption patterns. For example, a study by Listewnik and Formela emphasizes that HEMSs significantly reduce electricity costs while improving energy efficiency by employing a human–machine interface (HMI) to facilitate user interaction and control [98]. Furthermore, research by Jo et al. highlights that the integration of IoT platforms within HEMSs has amplified consumer interest in reducing energy costs, stressing the critical role of connectivity in modern energy management [173].

Studies indicate a growing body of research on energy-efficient appliances, particularly in developing regions like the Asia–Pacific [174]. These appliances offer significant benefits, such as lower energy bills by optimizing energy consumption, and a reduced environmental impact by minimizing greenhouse gas emissions and reliance on fossil fuels, contributing to a more sustainable energy system. However, barriers to widespread adoption persist, including high initial costs and limited consumer awareness [175]. The Theory of Planned Behavior, which posits that individual intentions are shaped by attitudes, subjective norms, and perceived behavioral control, has been applied to assess consumer intentions regarding the purchase of energy-efficient appliances. Within this framework, both perceived behavioral control and subjective norms have been found to exert a significantly positive influence on purchasing decisions [176]. A meta-analysis encompassing 30 studies identified attitude as the most influential factor in purchase intention, followed by perceived behavioral control and subjective norms [177]. These findings highlight the ongoing need for strategic policy interventions, technological innovations, and consumer education to foster greater adoption of energy-efficient appliances. Sarikaya and Partal highlight the significance of smart grids and advanced devices in enabling a two-way flow of energy, potentially reducing overall consumption by up to 20% [178]. This integration not only allows smart homes to operate more efficiently but also enables them to contribute excess energy back to the grid, enhancing the sustainability of the broader energy system. Additionally, research by Wang and Abdalla explores the optimization of energy scheduling in smart homes that incorporate PV systems and energy storage, further emphasizing the benefits of renewable energy integration [179].

Beyond technological advancements, the design and architecture of smart home systems are crucial to maximizing energy efficiency. Kurniawan’s research proposes a smart home architecture that prioritizes energy conservation and the management of multiple energy sources, utilizing fuzzy multi-criteria decision-making models and artificial neural networks [180]. This approach not only facilitates energy savings but also increases user awareness of energy consumption, encouraging more sustainable behavior. Moreover, Andreadou et al. examine the interoperability of smart home automation systems under demand response schemes, emphasizing the need for cohesive energy management strategies that dynamically adapt to fluctuating energy demands [181]. The economic feasibility of investing in smart home technologies is another important consideration. Larionova’s analysis provides insights into the long-term financial benefits of adopting smart home systems, signifying that initial investments are justified by the energy savings and operational efficiencies realized over time [182]. This cost–benefit perspective is critical in promoting the widespread adoption of energy-efficient technologies in residential settings.

Furthermore, this emphasis on user-centric design is echoed in the work of Al-Naima, who introduces a prototype smart home energy management system that aligns with both user comfort and cost criteria [183]. Such systems not only enhance energy efficiency but also improve the overall user experience by offering tailored solutions. The convergence of smart home technologies with emerging fields such as blockchain and edge computing further amplifies the potential for energy-efficient solutions. Iqbal et al. explore a blockchain-based edge-computing approach for smart home monitoring, which secures IoT devices and predicts energy usage, ensuring both efficiency and privacy [184]. This innovative integration reflects the ongoing evolution of smart home systems and their ability to address new challenges in energy management. These developments indicate that the future of smart homes lies in the seamless integration of advanced technologies, user-centered design, and renewable energy systems to create more efficient, sustainable, and economically viable energy solutions. Some applications of energy-efficient appliances in smart homes and their integration are summarized as follows:

• Energy-Efficient Lighting: LED smart bulbs, such as Philips Hue [185], offer energy-efficient lighting solutions that are controlled remotely. These bulbs consume significantly less energy than traditional incandescent bulbs and have a longer lifespan. A study by Lee and Kim demonstrated that integrating smart lighting with occupancy sensors reduces energy usage by up to 30% in residential settings [186]. This integration of smart lighting systems with motion sensors and home automation platforms allows lights to turn off automatically when no one is present, further enhancing energy savings.
• Smart Appliances: Modern smart appliances, such as refrigerators, washing machines, and dishwashers, are designed to be energy-efficient and controlled via smartphone applications. These appliances often feature energy-saving modes that automatically adjust their operation based on energy availability and demand. For instance, smart washing machines know how to schedule cycles during off-peak hours when electricity rates are lower, contributing to overall energy savings [187]. Furthermore, the implementation of ML algorithms in these appliances allows for predictive maintenance and optimized energy use, enhancing their efficiency [188].
• HEMSs are integral to the smart home ecosystem, allowing for the centralized control of various energy-efficient appliances. These systems monitor energy consumption in real time and provide insights into usage patterns, enabling homeowners to make informed decisions about their energy consumption [189]. By integrating renewable energy sources, such as solar panels, with HEMSs, homeowners are able to further enhance their energy efficiency and sustainability [190,191].
• Energy Monitoring Systems: Smart homes equipped with energy monitoring systems allow homeowners to track their energy usage in real time. This capability enables users to identify high consumption patterns and make adjustments to reduce waste. Continuous data collection helps optimize energy use patterns and leads to significant cost savings on utility bills ranging from 5% to 22% [18].

4.4. IoT, Blockchain, and Decentralized Systems in Smart Home Management

Recent research highlights the transformative potential of emerging technologies in optimizing smart home energy management, one of which is IoT. IoT plays a fundamental role by enabling real-time monitoring of both energy production and consumption, providing valuable data for system optimization [192]. When integrated with AI and big data analytics, these technologies allow for the prediction of consumption patterns, fine-tuning energy use, and enhancing overall system efficiency [193]. Additionally, integrating the IoT–blockchain framework for urban energy systems combines hybrid consensus (PoS + PBFT), K-means clustering for demand-response optimization, and lightweight protocols (MQTT/CoAP) to achieve energy efficiency, scalability, and user adoption, delivering a 15% energy cost reduction for high-consumption clusters, 80% lower energy use per transaction compared to Proof of Work, and near-linear scalability for 500+ IoT devices [194]. The development of smart home architectures that prioritize energy conservation, combined with the management of multiple energy sources, enables homeowners to control energy usage conveniently via smartphones or personal computers [180].

Concurrently, blockchain technology facilitates P2P energy trading and bolsters data security, creating more resilient and transparent energy networks [195]. Blockchain technology is increasingly recognized as a transformative component in smart home energy management. Its decentralized approach to energy transactions enhances the security and transparency of energy trading among prosumers (individuals who both consume and produce energy) [184]. This innovation removes the need for intermediaries, enhancing the efficiency of energy exchanges [196]. Various models have been proposed to optimize these transactions, including fully P2P trading strategies that leverage surplus matching and geographical proximity, as reported by Alskaif et al. [197], as well as auction-based platforms integrating IoT and blockchain technologies [198]. Ethereum-based smart contracts are widely adopted to ensure secure, transparent, and automated energy trading processes [199]. These systems provide several advantages, including reduced energy procurement costs, minimized reliance on centralized grids, and increased financial returns for participants [197]. However, blockchain-enabled P2P energy trading systems present a promising avenue for enhancing efficiency, security, and decentralization in smart grid energy management. The convergence of IoT, ML, blockchain, and optimization technologies establishes a robust framework for managing energy consumption in smart homes, which results in reduced carbon footprints and improved user comfort.

5. Optimization Methods Used in Energy Management of Smart Homes

While several reviews have explored artificial intelligence applications in broader renewable energy domains such as smart grids, wind farms, and energy storage systems [200], fewer have focused on how these technologies are specifically adapted for smart home energy optimization. Building upon such foundational work, this section reviews key AI optimization methods used in managing energy within residential environments.

Optimization methods are integral to the effective management of energy in smart homes, balancing the goals of maximizing user comfort while minimizing both energy consumption and costs [201]. Recent studies have explored diverse approaches to achieving these objectives. Liu highlights that advances in smart home technologies enable real-time monitoring of environmental conditions through sensors, allowing intelligent energy management that reduces household energy consumption, lowers carbon emissions, and supports sustainable development [202]. Additionally, advanced optimization techniques, such as the enhanced northern goshawk optimization (ENGO) algorithm, have demonstrated superior performance in concurrently reducing peak-to-average ratios, lowering electricity costs, and enhancing user comfort [203]. Meta-heuristic load scheduling strategies have also shown promise, with potential electricity cost reductions of up to 4.5% [204]. Moreover, incorporating weather-related variables, such as air pressure, dew point, and wind speed, into optimization models has proven effective in balancing energy savings with user convenience [187]. These advancements in optimization strategies not only enhance the efficiency of smart homes but also contribute to sustainable development and reduced environmental impact. These methods are applied to key aspects such as scheduling household appliances, managing energy storage, and integrating RESs [205]. Among the various optimization approaches, mathematical models, meta-heuristic techniques, and AI-based methods are commonly employed [206]. Notably, algorithms like genetic algorithms (GAs) and PSO have demonstrated considerable success in improving user comfort while simultaneously reducing energy consumption [207]. Shirsat et al. optimized price-based response programs using GAs and PSO to schedule smart home loads and manage IoT energy use. This approach reduces energy costs, enhances user comfort, and promotes sustainability in smart homes, supporting global energy management [208]. Sameh Mahjoub et al. enhanced energy management systems using Long Short-Term Memory neural networks optimized with GAs and PSO. Tests on real datasets showed improved prediction accuracy, with correlation coefficients ranging from 99.16% to 99.97%, highlighting the methods’ efficiency [209]. The central objectives of power scheduling problems in smart homes (PSPSH) focus on minimizing electricity bills, ensuring the stability of the power system, and optimizing user comfort by reducing appliance waiting times [210]. By achieving these goals, smart home systems offer both economic and environmental benefits while enhancing the overall user experience.

5.1. Comparison of AI Techniques and Traditional Methods in Smart Home Energy Optimization

Significant progress has been made in smart home energy management, where optimization algorithms, artificial intelligence (AI), and demand response strategies constitute the core methodological pillars; however, as residential energy demands become increasingly heterogeneous, achieving effective multi-objective coordination among load balancing, cost minimization, carbon reduction, user comfort, and demand response remains a major challenge. Mixed-integer linear programming (MILP) has emerged as a dominant approach for residential energy scheduling due to its capacity to model both discrete and continuous decision variables and its transparent optimization structure, yet its practical deployment is constrained by exponential computational complexity, slow convergence, and high computational cost, particularly under large-scale or real-time settings [211]. To address the limitations of conventional optimization frameworks, recent research has increasingly explored AI-driven and bio-inspired optimization paradigms. Makhadmeh et al. [212] examine bio-inspired optimization algorithms and their application to Natural Language Processing tasks, highlighting how evolutionary and swarm-based methods address optimization challenges in feature selection, parameter tuning, and model adaptability while identifying current limitations and future research directions. Recent evidence from high-impact studies (2023–2025) indicates that Generative AI architectures, particularly GAN-based and VAE-driven models, achieve 15–20% RMSE reductions in solar and wind forecasting and deliver 9–12% improvements in energy efficiency and curtailment reduction by effectively modeling nonlinear, uncertain dynamics in renewable-dominant power systems, thereby enhancing grid resilience and operational flexibility [213]. In addition, optimized integration of renewable energy sources and peak shaving using AI have led to a reduction in CO₂ emissions by 15%, compared to traditional methods. Furthermore, AI systems have reduced peak demand by 40%, improving grid stability and increasing self-consumption rates in homes with renewable energy systems [214]. At the appliance and household level, Islam et al. [215] focus on the methodologies, aims, and results of previous studies on smart home energy management, highlighting how machine learning models, particularly the Random Forest model, have been applied to predict and optimize appliance energy use, with results showing a 95.71% accuracy in appliance energy prediction and 99.71% accuracy for time series data, leading to significant improvements in energy efficiency and cost savings without compromising user comfort. In another study by Pirouz et al. [216], the literature on AI techniques for energy prediction highlights various methods such as Random Forest (RF), Long Short-Term Memory (LSTM), artificial neural networks (ANNs), and Extreme Gradient Boosting (XGBoost) as the most effective for predicting energy consumption and renewable energy production, with several studies comparing these techniques using performance metrics like Mean Absolute Error (MAE), Mean Squared Error (MSE), and R-squared (R²), with results indicating that RF and XGBoost consistently perform well across multiple datasets.

The primary advantages of AI over traditional methods in smart home energy management lie in its ability to process vast real-time data, deliver faster and more accurate decision-making through predictive analytics and automation, and overcome the limitations of static analysis and historical data reliance inherent in conventional systems, as evidenced by Rhali et al. [217], who highlight the advantages of AI over traditional methods in management control, emphasizing AI’s ability to process vast amounts of real-time data and enhance decision-making speed, accuracy, and agility through predictive analytics and automation, while overcoming the limitations of static analysis and historical data reliance found in conventional systems, with empirical research conducted in Morocco and abroad showing significant improvements in organizational performance. Lai’s survey of 75 participants also shows that 89% of respondents use at least one AI-powered smart home product, primarily for convenience and voice control, while adoption is constrained by product incompatibility (56%), system instability (50%), and privacy concerns (40%), with future expectations emphasizing greater personalization (77%) and integration of local community services (69%), highlighting AI’s potential to address current limitations and enhance smart home adoption and functionality [218]. Rehman et al. [39] conducted a review using the PRISMA protocol and examined energy management models for smart homes from 2018 to 2024, focusing on AI and machine learning-driven algorithms, energy optimization, renewable energy integration, and challenges such as data privacy and consumption variability, with findings showing advancements in energy efficiency, home automation, and grid stability, but highlighting the need for further research in demand-side management and AI/ML applications.

5.2. Specific Optimization Techniques Used in Smart Homes

Mixed-integer linear programming (MILP) is widely used for scheduling and load balancing in smart homes. It helps in optimizing energy consumption by determining the best times to operate appliances based on energy prices and demand as described in (1) and (2) [219].

$$Minimize C=\sum _{i=1}^n{c}_i{x}_i$$

(1)

Subject to:

$$\sum _{j=1}^m{a}_{i,j}{x}_j\le b_i {\forall }_i$$

(2)

where $C$ is the total cost (USD), ${\mathrm{c}}_{\mathrm{i}}$ is the cost associated with appliance $\mathrm{i}$ $\left(\mathrm{USD} \mathrm{per} \mathrm{unit}\right)$, ${\mathrm{x}}_{\mathrm{i}}$ is the binary decision variable indicating whether appliance $\mathrm{i}$ is on or off, ${\mathrm{a}}_{\mathrm{i},\mathrm{j}}$ represents the resource consumption, and ${\mathrm{b}}_{\mathrm{i}}$ is the resource availability constraint. F. Hasanlu et al. developed an MILP model for 24 h smart home energy scheduling, optimizing appliance performance and reducing costs. The model incorporates grid interaction, renewable resources, and storage. The results show a 45% reduction in grid power purchases, energy transfer to the grid, and a 65% decrease in the energy bill [220]. Gomes also introduced an advanced HEMS using MILP-based model predictive control (MPC); the MILP-based MPC approach yields superior results compared to MILP alone, as demonstrated in a real case study in Algarve, Portugal, utilizing updated information [221].

PSO, a computational method inspired by how birds flock or fish school in nature, is applied to energy prediction and optimization by adjusting appliance usage based on environmental factors like temperature and humidity. In simple terms, PSO helps find the best solution to a problem by mimicking the social behavior of animals that move together in a group [222], as described in (3) and (4).

$$v_i^{t+1}=w{v}_i^t+c_1{r}_1\left(p_i-x_i^t\right)+c_2{r}_2\left(g-x_i^t\right)$$

(3)

$$x_i^{t+1}=x_i^t+v_i^{t+1}$$

(4)

where ${\mathrm{v}}_{\mathrm{i}}^{\left(\mathrm{t}\right)}$ and ${\mathrm{x}}_{\mathrm{i}}^{\left(\mathrm{t}\right)}$ denote the velocity and position of particle $\mathrm{i}$ at iteration $\mathrm{t}$ with the same units as the decision variable, $\mathrm{w}$ is the dimensionless inertia weight, ${\mathrm{c}}_1$ and ${\mathrm{c}}_2$ are dimensionless cognitive and social acceleration coefficients, ${\mathrm{r}}_1$ and ${\mathrm{r}}_2$ are dimensionless random numbers in $[0,1]$, ${\mathrm{p}}_{\mathrm{i}}$ is the personal best position, $\mathrm{g}$ is the global best position, and $\mathrm{t}$ is the iteration index [187]. Kah Poh Lee et al. explored using PSO to optimize energy consumption in smart homes, focusing on managing non-shiftable appliances to prevent short blackouts. A mobile app was developed to monitor and optimize energy use based on human activities, with the results showing that PSO significantly improved energy consumption compared to other methods, suggesting its potential for broader application in real-world smart homes [223]. Menos-Aikateriniadis et al. reviewed PSO methods for scheduling and controlling residential energy resources, including smart appliances, electric vehicles, heating/cooling devices, and storage. They highlight PSO’s efficiency in resource scheduling and suggest future research for its broader application in real-world demand response systems [224].

Heba Youssef used an IBES optimization algorithm for home energy management in smart homes, aiming to reduce electricity costs, balance load demand, and enhance user comfort. The system coordinates appliances in real time and reschedules tasks using dynamic programming to minimize peak-to-average ratios and optimize energy use across different pricing strategies [225]. The Improved Bald Eagle Search (IBES) optimization algorithm focuses on optimizing energy usage in smart homes by managing load demand and scheduling appliances efficiently. The objective function can be defined as (5):

$$\min E_{total}=w_{1}C+w_{2}PAR+w_{3}WT$$

(5)

where $C$ denotes the total electricity cost, PAR is the peak-to-average ratio of the scheduled load, WT represents total appliance waiting time (user discomfort), and $w_1,w_2,w_3$ are weighting coefficients. The IBES optimization algorithm outperforms traditional methods with faster convergence, higher accuracy, better global and local search, efficient load balancing, and adaptability to dynamic energy conditions [226].

Ant colony optimization (ACO) for dynamic scheduling of home appliances in smart grids mimics the foraging behavior of ants, where pheromone trails guide the search for optimal paths. In smart home energy management, ACO optimally schedules appliances by shifting operations to off-peak hours, reducing costs and balancing grid demand [227]. Siregar et al. [228] applied metaheuristic optimization methods, specifically ant colony optimization and symbiotic organism search, to home energy management systems to optimize appliance scheduling and reduce electricity costs, and found that symbiotic organism search achieved cost reductions of up to 24.75%, outperforming traditional optimization approaches. The optimization can be expressed as (6):

$$p_{ij}={\displaystyle \frac{{\left[T_{ij}\right]}^{\alpha }{\left[n_{ij}\right]}^{\beta }}{\sum {\left[T_{ij}\right]}^{\alpha }{\left[T_{ij}\right]}^{\beta }}}$$

(6)

where $p_{ij}$ represents the probability of moving from node $i$ to node $j$, ${\tau }_{ij}$ denotes the amount of pheromone on the path between node $i$ and node $j$, $\alpha$ controls the influence of the pheromone intensity, ${\eta }_{ij}$ indicates the desirability of selecting the path from node $i$ to node $j$, and $\beta$ determines the relative importance of this desirability in the decision-making process. Degala Isaac proposed an optimization approach for energy scheduling in smart homes, aiming to reduce electricity costs and the peak-to-average ratio (PAR). The system integrates real-time electricity price (RTEP) with demand response and uses an ACO algorithm to ensure efficient appliance scheduling while preventing blackouts. The approach enhances grid stability by minimizing both electricity price and PAR [229]. Yadav, Ravindra Kumar et al. applied ACO to minimize day-ahead demand-side management (DSM) in areas with many appliances, including self-protection and self-organizing features. The study compares ACO with other algorithms, showing it outperforms Evolutionary Algorithm, Moth Flame Optimization, and Bacterial Foraging Optimization in reducing consumer energy costs and utility peak load [230].

Fuzzy logic control (FLC) enhances smart home energy management by handling uncertainty and making adaptive decisions. It improves energy efficiency by optimizing appliance usage in real time. FLC enhances user comfort by balancing energy savings with preferences. It supports efficient demand response by prioritizing essential loads. Unlike traditional methods, FLC effectively manages fluctuations in energy supply and demand. It also integrates seamlessly with AI and IoT for smarter automation [231,232]. This study applies FL to manage energy in smart homes, accounting for uncertainties in energy demand and supply [233]. The total energy consumption $E{C}^{o total }$ is as seen in Equation (7)

$$E{C}^{o total }=\sum _{aϵA}{y}_a^{h=1}$$

(7)

where $E{C}^{o total }$ denotes the total energy consumption (kWh), ${\mathrm{y}}_{\mathrm{a}}^{(\mathrm{h}=1)}$ represents the energy consumed by appliance $\mathrm{a}$ during hour $\mathrm{h}=1$ (kWh), and $\mathrm{A}$ is the set of considered appliances.

Beheshtikhoo et al. proposed type-2 fuzzy logic controllers for demand-side energy management in smart homes, integrating renewable sources and energy storage. The system reduces grid electricity consumption by 49.186 kWh daily and 343.95 kWh weekly, cutting electricity costs by 71.5% and lowering the peak-to-average ratio by 64.6% [234]. Dimitrios Kontogiannis et al. developed a fuzzy control system that uses environmental data and ML to optimize energy consumption in residential buildings. The system employs the Mamdani approach and decision tree linearization for rule generation, with feature selection based on XGBoost. It reduces the rule set size and improves computation speed, offering more accurate energy consumption recommendations [235].

The genetic algorithm-based optimized energy management for smart homes is a model that prioritizes user comfort and profitability in smart home energy management. It emphasizes the integration of renewable energy sources and the dynamic handling of energy consumption patterns using GAs, showcasing the effectiveness of the proposed optimization strategies [236], as in Equation (8).

$$Z\text{-}score\left(ACP\right)={\displaystyle \frac{ACP-\mu }{\sigma }}$$

(8)

where $\mathrm{Z}\text{-}\mathrm{score} \left(\mathrm{ACP}\right)$ is the standardized (dimensionless) value of ACP, $\mathrm{ACP}$ denotes the original parameter expressed in its physical units, $\mathsf{\mu }$ is the mean of ACP with the same units, and $\mathsf{\sigma }$ is the standard deviation of ACP with the same units.

Morteza et al. proposed a management protocol for optimal day-ahead scheduling of smart home appliances using sensors, fog nodes, and cloud processing. A neural network predicts user workload, while a genetic algorithm optimizes load factor and economic efficiency. Numerical results show improvements over conventional methods [237]. Obi et al. proposed an innovative multi-objective optimization framework combining digital twin technology with a hybrid neural network and the multi-objective genetic algorithm (HNN-MOGA), along with MADRL-driven dynamic pricing, to manage grid-connected village-level microgrids using solar and wind RES; Python-based simulations on the Shinhyocheon village MG in Gwangju, South Korea, achieved a profit of USD 910,789.16, an investment cost of USD 1,192,735.12, and a computation time of 8.59 s over 24 h [238].

5.3. Hybrid Optimization Techniques

Hybrid optimization techniques combine multiple algorithms or techniques to solve problems. They can be used in many fields, including compiler heuristics, ML, and energy management [239]. Recent advancements in smart home energy management have increasingly focused on hybrid optimization techniques to effectively balance energy efficiency and user comfort. El-Afifi et al. [240] introduced a multi-objective approach for demand-side management, achieving substantial reductions in both peak-to-average ratio and electricity costs. Similarly, they proposed a hybrid method combining the grey wolf optimizer and PSO, incorporating additional weather metrics to enhance energy prediction and optimization accuracy. Rajesh et al. [241] developed a hybrid Sailfish Optimizer and Adaptive Neuro-Fuzzy Inference System to optimize power management in IoT-enabled smart homes. Xu et al. [242] presented a decentralized federated learning-based method for predictive energy optimization, utilizing deep learning models alongside PSO, resulting in a 38% reduction in energy consumption compared to competing approaches. These studies highlight the significant potential of hybrid optimization techniques to enhance both energy efficiency and user comfort in smart home environments. Nasrollahzadeh et al. proposed a hybrid whale optimization algorithm (WOA) and PSO for optimized motion sensor placement in smart homes. The combination overcomes PSO’s limitations, offering improved coverage, detection accuracy, and reduced operating costs [243].

5.4. Comparative Analysis of Optimization Methods for Smart Home Energy Management

Efficient energy management in smart homes is essential for reducing energy costs, optimizing consumption, and enhancing sustainability under dynamic pricing and demand response conditions, motivating the development of advanced optimization-based approaches to address the growing complexity of residential energy systems integrating distributed energy resources and grid interaction [244]. This comparative analysis evaluates the strengths, weaknesses, and use cases of several prominent optimization techniques, including ACO, GA, PSO, MILP, FLS, and WOA [245]. Each method offers distinct advantages, such as flexibility, computational efficiency, and accuracy in load scheduling, demand-side management, and renewable energy integration. However, the selection of an appropriate optimization method depends heavily on contextual factors, such as home size, appliance types, and energy tariffs [246,247]. Hybrid models, combining elements from different optimization algorithms, show promise for addressing the challenges of scalability and dynamic adaptation.

Table 4. Summary of AI-based optimization methods for smart home energy management.

Ref	Method	Strengths	Weaknesses	Use Cases	Cost Reduction Rate (%)	Carbon Emission Reduction (%)	Computation Time/Complexity	Key Experimental Conditions/Notes
[229]	Ant Colony Optimization (ACO)	Effective for dynamic scheduling; balances energy costs with comfort	Computationally intensive; may require fine-tuning for optimal performance	Scheduling of home appliances, especially in smart grids with real-time pricing	~2.2 (monthly bill: unscheduled USD 217.88 → ACO USD 213.05)	N/R	N/R	$\mathrm{Residential} \mathrm{smart} \mathrm{home}; \mathrm{shiftable} \mathrm{appliances}; 24 \mathrm{h} \mathrm{RTEP} + \mathrm{IBR}; \mathrm{baselines}: \mathrm{unscheduled} (217.88, PAR1.81) \mathrm{v}\mathrm{s}.$ $\mathrm{d}\mathrm{e}\mathrm{t}\mathrm{e}\mathrm{r}\mathrm{m}\mathrm{i}\mathrm{n}\mathrm{i}\mathrm{s}\mathrm{t}\mathrm{i}\mathrm{c} (207.95$, PAR 1.95); ACO balances (USD 213.05, PAR 1.60); no renewables; nonlinear problem. a
[248]	Multi-objective Genetic Algorithm (MOGA)	Good for balancing multiple objectives; robust optimization	Can be slow to converge; high computational cost for large problems	Load scheduling, demand-side management, and integrating renewable energy	N/R	N/R	N/R	Microgrid (MG) with renewable energies (e.g., wind as compromise between cost/pollution); demand response (DR) programs, reactive loads, reserve scheduling; uncertainties in renewables/load; baselines: no DR/reactive loads (higher generation/reservation/startup costs and pollution); key result: 16% reduction in reservation costs with DR participation; multi-objective: minimize cost + GHG emissions; stochastic programming; no exact overall cost % or emission % reported. b
[249]	Improved Particle Swarm Optimization (XPSO)	Simple implementation; good for continuous optimization problems	Can get stuck in local minima without proper tuning	Appliance scheduling based on environmental factors (e.g., temperature, humidity)	N/R	N/R	N/R	Steel slab temperature prediction in reheating furnace; simulation data sets (random furnace temps 1000–1300 °C, billet sizes); measured data sets from actual furnace; baselines: PSO, IPSO, IPSO2, HPSO, CPSO (for benchmarks); WOA, IA, GWO, DE, ABC (for prediction); metrics: MAE < 1 °C, RMSE < 1.2 °C for XPSO; focus on accuracy/robustness, not energy, no cost/emission data reported. c
[250]	Mixed-Integer Linear Programming (MILP)	Precise optimization for cost reduction and load balancing	Computationally expensive; less suitable for real-time applications	Energy scheduling in homes with diverse appliances and energy tariffs	10.2 (only grid); 46.4 (grid + solar); 21.4 (grid + solar + storage with discharge)	N/R	<10 s/high (exact solver for MILP)	Residential smart home; 7 shiftable appliances; day-ahead RTP from Indian energy exchange; 24 h horizon; user time preferences/energy requirements; scenarios: only grid, grid + 1 kW PV, grid + PV + 0.5 kW storage (charge/discharge); baselines: unoptimized (cost 117.86 Rs, PAR 2.5062); results: optimized costs/PAR reductions as above; multi-objective (cost + PAR minimization); no carbon data reported. d
[251]	Fuzzy Logic System (FLS)	Handles uncertainty well; adaptable for real-world scenarios	Requires expert knowledge for system design; less effective in highly dynamic environments	Energy management with uncertain demand patterns and supply conditions	N/R	N/R	N/R	Smart grids with renewables (solar, wind, biomass); simulated data for performance metrics; real-world validation from operational installations; baselines: traditional PID controllers; results: 20% increase in renewable consumption, 15% decrease in grid frequency variations (stability), 25% enhancement in energy storage SOC (reliability), 22% overall system efficiency improvement; vs. PID: 10% less frequency deviations, 15% better SOC, 12% efficiency boost; sensitivity analysis shows robustness to parameter variations; focus on grid stability, renewable integration, and sustainability. e
[252]	Improved Genetic Whale Optimization Algorithm (WOA)	Excellent for load scheduling and minimizing energy costs	Requires parameter tuning; may not be as flexible as other methods	Scheduling in residential buildings with renewable energy sources	Costs reduced by ~92.896 yuan/day average	Emissions reduced by ~0.091 tons/day	N/R	Building-integrated energy system (gas turbines, wind/solar, ground source heat pumps, EV, central air-conditioning, energy storage); regional complex building evaluation; multi-objective (economic efficiency + minimal carbon emissions); pre-processed regional data; baselines: traditional scheduling (higher costs/emissions); optimized balances economy/environment; promotes “zero scenery waste”; no specific computation details reported. f
[195]	Hybrid Grey Wolf Optimizer and PSO	Incorporates weather metrics for improved accuracy	Complexity in integration with multiple factors	Energy prediction and optimization	Up to 30 (savings through energy trading over original expenses)	Minimized environmental impact	Transaction processing: ~1 s per transaction; confirmation immediate; throughput ~95%	Smart homes with solar panels and wind turbines; datasets: 20/50/100/200 houses over 365 days + Saudi Arabia weather-based case study (50 houses); real-time data collection via IoT; EWMA for consumption/production prediction; P2P trading via blockchain/smart contracts; baselines: traditional grid management (higher costs, no trading; revenue/costs compared); results: reduced energy costs, positive net revenue/savings; secure transactions. g

Table notes: ^a Simulated residential smart home with shiftable appliances and user-defined preferences; baselines: unscheduled and deterministic scheduling; inconsistencies: no renewable integration or carbon metrics reported; absolute cost values with qualitative PAR reduction (~11.6%); limits direct percentage-based comparisons. ^b Stochastic microgrid incorporating demand response, reactive loads, and renewable uncertainties; baselines: absence of demand response (higher reservation and pollution costs); inconsistencies: 16% reservation cost reduction only (no overall cost or quantitative GHG percentage); microgrid-scale results may not translate directly to residential contexts. ^c Industrial reheating furnace temperature prediction using simulated and measured data; baselines: standard PSO variants and other metaheuristics (WOA, GWO, DE, ABC); inconsistencies: non-energy-management focus (prediction accuracy only); no cost or carbon indicators; iteration counts (500–8000) not comparable to real-time energy scheduling. ^d Residential smart home with day-ahead real-time pricing, photovoltaic, and storage scenarios; baselines: unoptimized loads (specific cost and PAR values); inconsistencies: reductions highly scenario-dependent (10.2–46.4%); no carbon emissions reported; limited appliance number restricts scalability inferences. ^e Simulated and real-world smart grid with solar, wind, and biomass integration; baselines: conventional PID control; inconsistencies: reported improvements in efficiency (22%), renewable utilization (20%), and storage reliability (25%) rather than direct cost/carbon reductions; qualitative GHG minimization; robustness demonstrated via sensitivity analysis. ^f Regional building-integrated energy system (gas turbines, renewables, heat pumps, EV, storage); baselines: conventional grid procurement; inconsistencies: absolute daily reductions (≈92.896 yuan cost, ≈0.091 tons emissions) without percentage baselines; multi-objective design; regional data may limit generalizability. ^g Multi-residential setups (20–200 houses) with solar/wind, IoT monitoring, and blockchain-enabled P2P trading; baselines: traditional grid without trading; inconsistencies: up to 30% savings specific to trading; qualitative environmental footprint reduction; blockchain throughput metrics do not reflect full optimization time; scale variations impact comparability.

summarizes these methods, providing a clear comparison to guide their application in real-world smart home systems.

5.5. Statistical Comparison of the Studies in Table 4

This subsection presents a comprehensive statistical comparison of experimental results across few optimization approaches, focusing on their performance in smart home energy management systems. The study by Isaac and Kumar [229] provides a statistical comparison of the experimental results across the three optimization approaches, unscheduled, deterministic, and ACO, offering valuable insights into their performance concerning electricity price and peak-to-average ratio (PAR). These two key metrics are critical in assessing the efficiency of energy management in smart homes, with a focus on both cost-effectiveness and grid stability. The results from this study offer clear comparisons of the effectiveness of each method, revealing their strengths and trade-offs. The electricity price analysis indicates that the average price across all three methods was USD 212.96 (SD = USD 4.97). The unscheduled approach resulted in the highest cost, which points out the inefficiency of not implementing any optimization strategy. Without optimization, energy consumption during peak periods leads to increased electricity costs. In contrast, the ACO and deterministic approaches yielded relatively similar prices, highlighting their effectiveness in reducing operational costs. The slight difference in cost between these two methods suggests that both approaches optimize electricity consumption effectively, but the ACO method, while slightly more expensive, compensates for this by enhancing other critical aspects, such as grid stability. The analysis of the PAR reveals a mean value of 1.7843 (SD = 0.175), reflecting a moderate level of fluctuation between peak and average demand. However, the ACO approach demonstrated the best performance, reducing the PAR to 1.5988, indicating a superior ability to flatten the demand curve and reduce peak loads. This reduction is crucial for improving grid stability, as it minimizes the stress on the grid during peak hours, thereby reducing the likelihood of blackouts and improving overall system reliability. The better performance of the ACO method in terms of PAR optimization demonstrates its effectiveness in managing demand-side energy and enhancing the overall efficiency of the energy management system. The small standard deviations for both electricity price and PAR suggest that the performance of these optimization methods is consistent and reliable across different trials. This consistency is vital, as it indicates that the results are not highly sensitive to variations in input conditions, reinforcing the robustness of the methods in real-world applications. However, the findings suggest that while the ACO approach has a slightly higher electricity cost, it offers substantial benefits in terms of grid stability by effectively minimizing the PAR. This ability to optimize grid stability is critical for minimizing blackouts and optimizing smart home energy management systems. The results confirm that ACO provides a balanced solution, ensuring both cost-effectiveness and grid stability. These advantages make ACO a promising method for energy management in smart homes, particularly in systems where grid stability and peak load reduction are essential considerations.

A statistical analysis of the experimental data of the study by Torkan et al. [248] reveals the following key findings summarized in Table 4. First, the analysis of production costs, reservation costs, and pollution levels in the microgrid system reveals important insights into the dynamics of energy generation, particularly with respect to the integration of renewable sources and the reliance on fossil fuels. The observed fluctuations in production costs throughout the 24 h period emphasize the significant impact of variable renewable energy generation on overall system expenses. The average production cost was found to be USD 34.13 (SD = USD 23.67), with a notable peak reaching USD 95.28 during hours when non-renewable generation sources, such as diesel generators, were required to meet demand. This variation suggests that while renewable energy sources like wind and solar are less expensive, the intermittency of these sources necessitates the use of more expensive fossil fuel-based power to stabilize the grid. The reliance on non-renewable generation thus introduces volatility into the cost structure, which could be mitigated through improved energy storage solutions or enhanced integration of renewables. Similarly, the reservation cost exhibited variability, with a mean of USD 1.09 (SD = USD 1.14), reflecting the fluctuating demand for reserve power to maintain grid stability. Higher reservation costs were particularly observed during periods of low renewable energy generation, where fossil fuel-based units were needed to ensure system reliability. The observed range from USD 0.16 to USD 4.35 in reservation costs further highlights the complexity of grid management, where maintaining reserves can become costlier when relying on traditional energy sources, especially during times of renewable intermittency. From an environmental perspective, the results emphasize the significant pollution levels associated with the use of fossil fuels. The average pollution level was found to be 48.68 kg/kWh (SD = 10.01 kg/kWh), with the highest pollution recorded at 63.92 kg/kWh during periods when diesel generators were used. These findings highlight the substantial environmental costs of maintaining grid stability through non-renewable sources, particularly during periods of low renewable generation. The contrast with renewable sources, which exhibited negligible pollution levels, further emphasizes the environmental benefits of increasing renewable penetration in the energy mix. By reducing reliance on diesel and other fossil fuels, the system could not only lower operational costs but also significantly reduce its carbon footprint. Overall, the data suggests that integrating RESs, such as wind and solar, can reduce both production and reservation costs while minimizing pollution. The relatively low standard deviations for both production cost (SD = USD 23.67) and reservation cost (SD = USD 1.14) indicate that the performance of the optimization method remains consistent across various scenarios. This stability supports the reliability of the optimization approach for energy management in microgrids, suggesting that such systems could be scalable and adaptable to varying levels of renewable energy generation. However, the results demonstrate the potential of renewable energy to significantly improve the economic and environmental performance of microgrid systems. However, challenges related to intermittency remain, pointing out the importance of investing in energy storage technologies and flexible grid management strategies. These findings provide valuable insights for future research and policy, reinforcing the need for continued development of technologies that enhance the integration of renewables while minimizing the costs and environmental impact of energy generation.

According to the statistical analysis in the study by Kesarkar and Priolkar [250], the experimental results of various load scheduling scenarios, optimized using MILP, offer valuable insights into the effectiveness of this method in reducing electricity costs and managing the PAR. A comprehensive cost analysis revealed an average electricity cost of 96.17 Rs (SD = 20.59 Rs) across all scenarios, which reflects moderate variation in the costs associated with different scheduling methods. The unoptimized scenario, which lacked MILP-based optimization, resulted in the highest cost of 117.86 Rs. This finding highlights the inefficiencies in energy consumption when dynamic scheduling based on real-time pricing signals is not applied. Without optimization, energy demand peaks during high-cost periods, driving up overall electricity expenses. Conversely, the lowest electricity cost of 63.14 Rs was achieved when both grid and solar power were integrated with the MILP-based optimization method. This reduction in cost emphasizes the significant role of renewable energy sources in mitigating electricity costs, particularly when coupled with optimization techniques. The variation in cost across scenarios, with a range of 54.72 Rs, illustrates the tangible benefits of employing MILP-based optimization to balance energy consumption more efficiently. These findings reinforce the potential for renewable energy integration, such as solar power, to lower costs, particularly when optimized through an intelligent scheduling approach. Further analysis of the peak-to-average ratio (PAR) across the scheduling scenarios showed a mean PAR of 2.04 (SD = 0.67), indicating moderate variation in peak load relative to average demand. The unoptimized scenario exhibited the highest PAR of 2.51, which emphasizes the strain on the grid when peak demand remains high due to the lack of optimization. Without optimization, the residential energy system is unable to shift energy consumption away from peak periods, increasing reliance on costly and less environmentally friendly grid power. However, the Grid + Solar + Storage (with discharge) scenario, optimized through MILP, achieved the lowest PAR of 1.23, demonstrating the effectiveness of combining renewable energy sources with energy storage systems in reducing peak loads. The significant reduction in PAR (with a range of 1.69) further highlights the role of MILP in balancing energy demand, not only by minimizing costs but also by mitigating grid stress during peak periods. This result is crucial, as it points to the advantages of integrating both renewable energy and storage technologies, which together improve the overall efficiency of residential energy systems. The results from the statistical analysis validate MILP-based optimization as an effective tool for reducing both electricity costs and the peak-to-average ratio (PAR) in residential energy systems. The unoptimized scenario demonstrated the highest cost and PAR, highlighting the importance of optimization in residential energy management. The best outcomes were observed when grid power was combined with solar energy, further enhanced by storage discharge, which contributed to a more balanced energy profile and reduced peak demand. These findings confirm that MILP can effectively balance cost minimization and peak load reduction, offering significant benefits to both consumers and utilities, particularly when coupled with renewable energy resources like solar power and energy storage systems. This analysis supports the integration of advanced optimization methods and renewable technologies as a pathway to more efficient and sustainable residential energy management.

In Usanova et al.’s [251] study, the experimental analysis of the fuzzy logic-based energy management model demonstrates its significant potential in improving key metrics such as renewable energy usage, grid stability, energy storage reliability, and overall system efficiency. The findings reveal that the model delivers substantial improvements across various system components, with consistent performance in enhancing renewable energy integration and optimizing grid management. These improvements are vital for advancing energy systems that depend on renewable sources, where stability and efficiency are critical. One of the most notable outcomes of the study is the improvement in renewable energy usage, which showed a mean increase of 20%, with a standard deviation of 5.8%. This improvement reflects the model’s ability to integrate renewable energy more effectively into the grid. The variability in performance, with improvements ranging from 10% in grid frequency deviation reduction scenarios to 25% in energy storage reliability, highlights the adaptable nature of the fuzzy logic model across different system components. Despite the variability, the model consistently contributes to enhancing renewable energy usage, confirming its potential to optimize renewable integration in real-world applications. The model’s influence on grid frequency variation also proves substantial. A mean reduction of 15% in frequency fluctuations was observed, with a standard deviation of 7.5%, indicating a moderate but reliable decrease in grid instability. The maximum improvement in grid frequency control was 15%, emphasizing the model’s ability to maintain grid stability, a critical factor for the successful integration of intermittent renewable energy sources such as wind and solar. This reduction in grid frequency variation not only enhances system stability but also helps avoid potential grid failures, making the fuzzy logic approach particularly valuable in systems with high renewable energy penetration. In terms of energy storage reliability, the fuzzy logic model demonstrated a remarkable mean improvement of 25% (SD = 5%), which is a significant enhancement compared to conventional methods. This result suggests that the fuzzy logic model optimizes the charging and discharging cycles of energy storage systems more efficiently, increasing their reliability. Such improvements in storage reliability are essential for managing the variability of renewable energy sources and ensuring that excess energy is efficiently stored for later use, reducing the dependency on non-renewable backup systems. Furthermore, the analysis of overall system efficiency revealed a mean improvement of 22% (SD = 6%), with the highest improvement observed in energy storage state of charge management (15%) and the lowest in comparison with traditional controllers (12%). These results reflect the superior efficiency of the fuzzy logic model, particularly in balancing the demands of energy storage and renewable integration. By optimizing various system components, the fuzzy logic model ensures that energy is used more effectively, leading to substantial gains in overall system performance.

The statistical analysis confirms that the fuzzy logic-based energy management model significantly outperforms traditional control strategies, such as PID controllers, in improving renewable energy utilization, grid stability, energy storage reliability, and overall system efficiency. The relatively low standard deviations observed across the metrics indicate that the fuzzy logic model is both robust and reliable in diverse operational scenarios. These findings highlight the value of fuzzy logic controllers (FLCs) in smart grid systems, particularly in scenarios where renewable energy integration is critical. The model not only enhances efficiency but also provides a more adaptable and stable framework for managing the complex dynamics of modern energy systems. Also, the results validate the fuzzy logic approach as a promising solution for optimizing energy management, especially in systems that rely heavily on renewable energy sources. By improving energy storage reliability, reducing grid frequency fluctuations, and increasing overall system efficiency, the fuzzy logic model offers a comprehensive strategy for enhancing the performance of smart grids. These improvements have the potential to pave the way for more sustainable, reliable, and efficient energy systems in the future.

In a study by Wei & An [252], the results of the experimental analysis conducted using the improved genetic whale optimization algorithm offer compelling evidence of its effectiveness in optimizing both economic costs and carbon emissions for building-integrated energy scheduling. The findings highlight the algorithm’s potential to achieve substantial improvements in both economic and environmental metrics, making it a valuable tool for optimizing energy usage in smart buildings. A detailed economic cost analysis revealed that the mean cost across the two algorithms was 3068.01 yuan (SD = 46.00 yuan). Algorithm 1: The improved genetic whale algorithm performs energy scheduling by optimizing the coordinated operation of gas turbines, wind power generation, and energy storage systems using demand and response data demonstrated superior performance, with a lower economic cost of 3021.557 yuan compared to Algorithm 2: The improved whale algorithm conducts energy scheduling using the same demand and response data as Algorithm 1, but without the genetic optimization component, which resulted in a cost of 3114.453 yuan. This 92.896-yuan difference between the two algorithms emphasizes the efficiency of Algorithm 1 in reducing operational expenses. The optimization achieved by Algorithm 1 highlights the algorithm’s capability to minimize energy-related costs in building operations, offering clear economic advantages for energy scheduling in real-world applications. Alongside the cost reduction, the analysis of carbon emissions further emphasizes the benefits of the improved genetic whale optimization algorithm. The mean carbon emissions across both algorithms were found to be 1.8355 tons (SD = 0.0455 tons). Algorithm 1 resulted in a carbon emission of 1.790 tons, while Algorithm 2 yielded a slightly higher 1.881 tons. The difference of 0.091 tons between the two algorithms illustrates a marginal increase in emissions with Algorithm 2. Despite both algorithms demonstrating relatively low emissions, Algorithm 1 outperforms Algorithm 2 in terms of minimizing the carbon footprint. This reduction in carbon emissions highlights the environmental benefits of using optimization algorithms to enhance energy scheduling, further reinforcing the value of Algorithm 1 in reducing both operational costs and environmental impact. The statistical analysis of both economic costs and carbon emissions confirms the effectiveness of the improved genetic whale optimization algorithm in optimizing building energy scheduling. Algorithm 1 stands out with its superior performance, not only achieving a significant reduction in daily operational costs by 92.896 yuan but also a notable decrease of 0.091 tons in carbon emissions. These results demonstrate the dual advantages of this optimization approach, ensuring both economic savings and environmental benefits. By optimizing energy consumption and reducing waste, the algorithm proves to be a promising solution for improving energy efficiency in buildings, with implications for broader applications in sustainable urban development. The findings substantiate the potential of the genetic whale optimization algorithm as a powerful tool for energy management in buildings, offering significant reductions in both operational costs and carbon emissions. The superior performance of Algorithm 1, particularly in minimizing both costs and environmental impact, showcases its effectiveness and provides strong evidence for its implementation in future smart building energy management systems.

5.6. Contextual Factors Influencing Method Selection

❖

Home Size: Larger homes may require more robust optimization techniques like MILP or GA, which can handle a broader set of appliances and more complex demand-response scenarios. Smaller homes might benefit from lighter methods like ACO or PSO that are computationally efficient [236].

❖

Appliance Types: Homes with a mix of high- and low-power appliances (e.g., HVAC and washing machines) may benefit from a method like MILP or FLS, which handles mixed energy demands. Homes with simpler, low-power appliances might use PSO or ACO for efficient scheduling with real-time energy tariffs [220,247].

❖

Energy Tariffs: For homes with dynamic pricing or demand-response programs, methods like GA or MILP, which can optimize energy consumption across periods and pricing levels, may be ideal. ACO can also be suitable for real-time optimization in such scenarios [253,254].

Table A3. Optimization of energy management in smart homes.

Ref	Summary	Optimization Methods Used	Result	Problem Formulation and Constraints
[97]	In this study, an optimized smart home energy management system (OSHEMS) was examined that ensured dependable load delivery while reducing grid reliance and energy costs.	The Home Energy Management Whale Optimization Algorithm (HEMWOA) was used.	Experimental results demonstrated a considerable decrease in grid reliance (46.6%) and energy expenses (57.7%) compared to non-scheduled scenarios.	Encircling the prey phase: $\overrightarrow{X}\left(t+1\right)=\overrightarrow{X^{\ast }}\left(t\right)-\overrightarrow{A}·\overrightarrow{D}$ $\mathrm{Distance} \mathrm{Definition} \overrightarrow{D}=|\overrightarrow{C}·\overrightarrow{X^{\ast }}\left(t\right)-\overrightarrow{X}\left(t\right)|$ Bubble net attacking phase: behavioral model Shrinking encircling prey model: decreasing the value of $\overrightarrow{a }$ which in turn leads to decreasing the value of $\overrightarrow{A}$ Spiral model $\overrightarrow{X}\left(t+1\right)=\overrightarrow{D}·e^{bl}·\cos \left(2\pi l\right)$ + $\overrightarrow{X^{\ast }}\left(t\right)$ $\overrightarrow{D}=|\overrightarrow{X^{\ast }} \left(t\right)-\overrightarrow{X}\left(t\right)|$ $\mathrm{Search} \mathrm{for} \mathrm{prey} \mathrm{phase} (\mathrm{exploration} \mathrm{phase}): \overrightarrow{X}(t+1)=\overrightarrow{X_{rand}}\left(t\right)-\overrightarrow{A}·\overrightarrow{D}$ $\overrightarrow{D}=|\overrightarrow{C}·\overrightarrow{X_{rand}}\left(t\right)- \overrightarrow{X}\left(t\right)|$ Home energy cost objective: $E_{Home}^T\left(t\right)={\displaystyle \sum _{t=1}^{24}}\left(\left({\displaystyle \sum _{s=1}^m}\left(E_{t,S}\left(t\right)\times W_{t,S}\left(t\right)\right)+{\displaystyle \sum _{N=1}^n}\left(E_{t,N}\left(t\right)\times W_{t.N}\left(t\right)\right) -E_{PV}\left(t\right) -E_B\left(t\right)\right)\times P\left(t\right)\right)$ where $\overrightarrow{\mathrm{X}}(\mathrm{t})$ and $\overrightarrow{\mathrm{X}}(\mathrm{t}+1)$ denote the current and updated whale positions at iteration $\mathrm{t}$, $, \overrightarrow{{\mathrm{X}}^{\ast }}(\mathrm{t})$ is the best solution found so far, ${\overrightarrow{\mathrm{X}}}_{\mathrm{r}\mathrm{a}\mathrm{n}\mathrm{d}}(\mathrm{t})$ is a randomly selected solution, $\overrightarrow{\mathrm{D}}$ is the distance vector, and $\overrightarrow{\mathrm{A}}$ and $\overrightarrow{\mathrm{C}}$ are dimensionless coefficient vectors controlling exploitation and exploration; $\mathrm{b}$ and $\mathrm{l}$ defines the spiral shape, and all position vectors are expressed in units consistent with the decision variables. The objective equation defines the total daily electricity cost of the smart home by accounting for flexible and non-flexible appliance energy consumption, on-site photovoltaic generation, battery contribution, and time-varying electricity prices over a 24 h scheduling horizon.
[274]	This paper framed the energy planning problem (EPP) as an optimization challenge aimed at determining the best schedules to reduce energy consumption costs and demand, while improving user comfort.	The grey wolf optimizer (GWO) was modified and tailored to solve the EPP in an optimal manner, efficiently achieving its goals.	To evaluate the effectiveness of the proposed method, its performance using the GWO with RESs was assessed in three phases: first, by comparing it with original methods without RESs; second, with methods that incorporate RESs; and third, by benchmarking against state-of-the-art approaches. The results demonstrated the method’s robust ability to tackle the EPP and optimize its objectives.	$\mathrm{Coefficient} \mathrm{vector} \overrightarrow{A}$: $\overrightarrow{GA}=2\times \overrightarrow{ga}\times {\overrightarrow{gr}}_1$ $- \overrightarrow{ga}$, $\mathrm{Coefficient} \mathrm{vector} \overrightarrow{C}$: $\overrightarrow{GC}=2\times \overrightarrow{{gr}_{2,}}$ $\mathrm{Distance} \mathrm{from} \mathrm{alpha} \mathrm{wolf} \overrightarrow{{GD}_{\propto }}=\left|\overrightarrow{G{C}_1}-\overrightarrow{G{X}_{\propto }}-\overrightarrow{GX}\right|$, $\mathrm{Alpha}\text{-}\mathrm{based} \mathrm{position} \overrightarrow{G{X}_1}=\overrightarrow{{GX}_{\propto }}-\overrightarrow{G{A}_1}\times \overrightarrow{{GD}_{\propto }},$ $\mathrm{Distance} \mathrm{from} \mathrm{beta} \mathrm{wolf} \overrightarrow{{GD}_{\beta }}=\left|\overrightarrow{G{C}_2}-\overrightarrow{G{X}_{\beta }}-\overrightarrow{GX}\right|$, $\mathrm{Beta}\text{-}\mathrm{based} \mathrm{position} \overrightarrow{G{X}_2}=\overrightarrow{{GX}_{\beta }}-\overrightarrow{G{A}_2}\times \overrightarrow{{GD}_{\beta }},$ $\mathrm{Distance} \mathrm{from} \mathrm{delta} \mathrm{wolf} \overrightarrow{{GD}_{\delta }}=\left|\overrightarrow{G{C}_3}-\overrightarrow{G{X}_{\delta }}-\overrightarrow{GX}\right|$, $\mathrm{Delta}\text{-}\mathrm{based} \mathrm{position} \overrightarrow{G{X}_3}=\overrightarrow{{GX}_{\delta }}-\overrightarrow{G{A}_3}\times \overrightarrow{{GD}_{\delta }},$ $\mathrm{Position} \mathrm{update} \mathrm{rule} \overrightarrow{GX}\left(itr+1\right)={\displaystyle \frac{\overrightarrow{{GX}_1}-\overrightarrow{G{X}_2}\times \overrightarrow{{GX}_3}}{3}},$ $\mathrm{Objective} (\mathrm{Fitness}) \mathrm{Function} minF\left(X\right)=w_1\times {\displaystyle \frac{EB}{EB+A}} +w_2\times {\displaystyle \frac{PAR}{PAR+B}} +w_3\times WT{R}_{avg}+w_4$ where $\overrightarrow{X}$ and $\overrightarrow{X}(\mathrm{itr}+1)$ denote the current and updated solution vectors at iteration itr, ${\overrightarrow{X}}_{\alpha }$, ${\overrightarrow{X}}_{\beta }$, and ${\overrightarrow{X}}_{\delta }$ are the positions of the three leading wolves, ${\overrightarrow{A}}_k$ and ${\overrightarrow{C}}_k$ are coefficient vectors controlling exploration and exploitation, ${\overrightarrow{r}}_1$ and ${\overrightarrow{r}}_2$ are random vectors uniformly distributed in $[0,1]$, and $\overrightarrow{a}$ is a linearly decreasing control parameter. The objective equation defines the multi-objective fitness function, where $EB$ is the energy bill, $PAR$ is the peak-to-average ratio, $WT{R}_{\mathrm{avg}}$ is the average waiting time ratio, $w_1$–$w_4$ are weighting coefficients, and $A$ and $B$ are normalization constants used to balance competing objectives in smart-home scheduling.
[275]	The study focused on sensor optimization in smart environments and aimed to improve activity recognition while managing power consumption and cost constraints. The method used spatial and temporal contexts represented by an ontology model for sensor mapping and activity recognition.	Sensor optimization was used.	The study focused on optimizing sensors in smart homes for activity recognition by using spatial and temporal data, removing inactive sensors, and pairing redundant ones while maintaining accuracy even in multi-resident environments with concurrent activities.	$\mathrm{Spatial} \mathrm{Noise} \mathrm{Elimination} pMI\left(l,a\right)=\log \left({\displaystyle \frac{p\left(l,a\right)}{p\left(l\right)p\left(a\right)}}\right)$ $\mathrm{Label} \mathrm{assignment}: l\left(a\right)=maxpMI\left(l,a\right)$ Using Mutual Information: $MI\left(sensor \left(S\right), Activity\left(A\right)\right)={\sum }_{iϵs}{\sum }_{aϵA}^{n}P\left(Si,a\right)\log \left[{\displaystyle \frac{P\left(S_{i, }a\right)}{P\left(S_i\right)P\left(a\right)}}\right]$ where $p(l,a)$ denotes the joint probability of sensor label $l$ and activity $a$, and $p\left(l\right)$ and $p\left(a\right)$ are their corresponding marginal probabilities; $\mathrm{p}\mathrm{M}\mathrm{I}(l,a)$ quantifies the strength of association between a sensor and an activity, $l\left(a\right)$ selects the most informative sensor label for activity $a$ $, S$ represents the set of sensors, and $A$ denotes the set of activities. The equations above collectively define a mutual-information-based spatial noise elimination strategy that identifies informative sensors while removing redundant or irrelevant sensor data.
[29]	The objective of the proposed system was to optimize the energy usage of SUB appliances to efficiently managed load demand. This leads to a reduction in the peak-to-average ratio (PAR) and a subsequent decrease in electricity costs, all while ensured that user comfort remained a top priority.	The improved sine cosine algorithm (ISCA) was evaluated against the grasshopper optimization algorithm (GOA).	This study compared the performance of the improved sine cosine algorithm (ISCA) with the grasshopper optimization algorithm (GOA) for optimizing energy consumption in smart urban buildings (SUBs). The proposed method achieves significant improvements in electricity cost, peak-to-average ratio (PAR), and waiting time, with reductions of 29.16%, 51.51%, and 35.07%, respectively. In comparison, GOA showed improvements of 13.72%, 38.00%, and 13.97%. The results demonstrated that ISCA outperforms both the unscheduled scenario and GOA, offering benefits for both utilities and consumers.	$\mathrm{Grasshopper} \mathrm{optimization} \mathrm{algorithm} \left(\mathrm{GOA}\right) P_i=S{O}_{i }+GR{E}_i+W_i$ $\mathrm{Weighted} \mathrm{position} \mathrm{update} P_i=r_{1}S{O}_{i }+r_{2}GR{E}_i+{r_{3}W}_i$ $\mathrm{Position} \mathrm{update} \mathrm{in} \mathrm{the} \mathrm{d}$-th dimension $P_i^d=c\left(\begin{array}{c}\sum _{j=1}^{N}c\\ j\ne i\end{array}{\displaystyle \frac{u{b}_d-l{b}_d}{2}}s\left(\left|P_j^d-P_j^d\right|\right){\displaystyle \frac{\left(P_j-P_i\right)}{\left|P_j-P_i\right|}}\right)+\widehat{T_d}$ $\mathrm{Velocity} \mathrm{update} \mathrm{rule} v_{i,t+1}=\{x_{i,j}+A\sin \left(r_1\right)\left|G{x}_{i,Best}-x_{i,t}\right|+r_2\left(x_{Best}-x_{i,t}\right)$ $<0.5x_{i,t}+Acos\left(r_1\right)\left|G{x}_{i,pBest}-x_{i,t}\right|+r_2\left(x_{Best}-x_{i,t}\right) O.W$ where ${\mathrm{P}}_i$ denotes the position vector of the $i$-th grasshopper, ${SO}_i$, ${GRE}_i$, and $W_i$ represent the social interaction, gravity, and wind-advection components, respectively; $r_1$, $r_2$, and $r_3$ are random weighting coefficients, $c$ is a decreasing control parameter, $u{b}_d$ and $l{b}_d$ are the upper and lower bounds of the $d$-th decision variable, $s(\cdot )$ is the social interaction function, and ${\widehat{T}}_d$ is the target position in dimension $d$. The velocity update rule equation introduces a velocity-based refinement mechanism that enhances convergence by adaptively balancing exploration and exploitation using global and personal best positions.
[276]	Specifically, the model permitted the integration of distributed renewable energy systems (DRESs) and takes TOU into account. Additionally, the cuckoo optimization technique is used to solve the load scheduling model. An equivalent MILP model to the load scheduling problem is created and solved in order to validate the cuckoo algorithm’s performance. Cuckoo and its equivalent MILP model are compared in terms of optimality and time performance through a series of exercises.	The cuckoo optimization technique was used for various operating conditions and scheduling criteria.	The cuckoo results, based on real data from an Egyptian academic building, demonstrated that under specific conditions, the building could achieve energy cost savings of 57% to 80%.	$\mathrm{Multi}\text{-}\mathrm{objective} \mathrm{fitness} \mathrm{function}:minf = w_{1}f 1 + w_{2}f 2 + w_{3}f 3$ Subject to: $\mathrm{Energy} \mathrm{consumption} \mathrm{equation} \mathrm{for} \mathrm{appliance} i$: $\sum _{t=1}^T{e}_{i,k}\left(t\right)=d_{i,k}·p_{i,k}, \forall i,k$ $\mathrm{Energy} \mathrm{consumption} \mathrm{bounds} \mathrm{for} \mathrm{appliance} \mathrm{i}$ $\mathrm{in} \mathrm{the} \mathrm{time} \mathrm{interval}: \sum _{t=1}^{v+d_{i,k}-1}{e}_{i,k}\left(t\right)=d_{i,k}·p_{i,k}, \forall i,k;v \in \left[{\tilde{L}}_{i,k}, {\tilde{U}}_{i,k}\right]$ $\mathrm{Lower} \mathrm{bound} \mathrm{of} \mathrm{energy} \mathrm{consumption}: {\tilde{L}}_{i,k}=max\left(1, r_{i,k}-X_{i,k}\right)$, $\forall i,k$ $\mathrm{Upper} \mathrm{bound} \mathrm{of} \mathrm{energy} \mathrm{consumption}: {\tilde{U}}_{i,k}=min\left(T-d_{i,k}+ 1, r_{i,k}+X_{i,k}\right),$ ∀i, k $\mathrm{Power} \mathrm{balance} \mathrm{constraint} \mathrm{at} \mathrm{time} t$: $N\left(t\right)+\sum _{k=1}^K\sum _{i=1}^{M_k}{e}_{i,k}\left(t\right)\le s\left(t\right)+PVG\left(t\right), \forall t$ $\mathrm{Final} \mathrm{optimization} \mathrm{goal}: {\displaystyle \genfrac{}{}{0pt}{}{min}{u}}$$f3=D$ where $f_1$, $f_2$, and $f_3$ denote individual objective functions associated with energy cost, peak load reduction, and user discomfort, respectively, and $w_1$, $w_2$, and $w_3$ are non-negative weighting coefficients satisfying $\sum w_i=1$. $e_{i,k}\left(t\right)$ represents the energy consumed by appliance $i$ of category $k$ at time $t$, $d_{i,k}$ is the required operating duration, and $p_{i,k}$ is the rated power. The scheduling window $[{\tilde{L}}_{i,k},{\tilde{U}}_{i,k}]$ is determined by the preferred start time $r_{i,k}$ and flexibility parameter $X_{i,k}$. $N\left(t\right)$ denotes non-schedulable load demand, $s\left(t\right)$ is the grid supply, $PVG\left(t\right)$ represents photovoltaic generation, and $D$ quantifies the aggregated discomfort metric minimized in the final optimization stage.
	The research proposed a new methodology for demand-side management and also used the Support Vector Regression technique to forecast a dispersed generation for the next day. The K-means clustering technique was used to identify the user comfort levels, which were validated by numerical simulations using actual data from a smart house.	Elite Non-dominated Sorting Genetic Algorithm II was used.	When comparing smart homes with and without distributed generation and battery banks, the effectiveness of the suggested AI combination was demonstrated by a 51.4% cost decrease.	$\mathrm{Photovoltaic} \mathrm{panel} \mathrm{power} \mathrm{equation}: P_w=\{\begin{array}{c}0\\ \left(a\times V^9\right)\\ 0\end{array}+\begin{array}{ccc}0& V& <V_i\\ \dots & +& \left(i\times V\right)+j, V_i\le V<V_0\\ V& \ge & V_0\end{array}$ $\mathrm{Power} \mathrm{of} \mathrm{the} \mathrm{photovoltaic} \mathrm{system}: P_p={\displaystyle \frac{N\times FF\times V}{1}}$ $\mathrm{For} \mathrm{N} \mathrm{the} \mathrm{Number} \mathrm{of} \mathrm{photovoltaic} \mathrm{panels} \mathrm{FF} ={\displaystyle \frac{V_{mppt}-I_{mppt}}{V_{oc}{-I}_{sc}}}$ $\mathrm{Optimization} \mathrm{Function} \left(\mathrm{Alg}30\right) \_i\in \{\mathrm{1,2},\dots ,├ k\},\left[f\right(x\_i )\ge f(y\_i \left)\right]^\left[{\exists }_iϵ \mathrm{1,2}\dots ,k:f\left(x_i\right)\right]$ $\mathrm{Objective} \mathrm{function} \mathrm{dominance}: P_s$:$\left\{x,y\in X|\exists F\left(y\right)>F\left(x\right)\right\}.$ $\mathrm{Set} \mathrm{of} \mathrm{optimal} \mathrm{solutions}: P_s:=\left\{F\left(x\right)|x\in P_s\right\}$ where $P_w\left(V\right)$ denotes the voltage-dependent power output of a photovoltaic panel, $V_i$ and $V_0$ are the cut-in and open-circuit voltages, respectively, and $i$ and $j$ are linear approximation coefficients. $P_p$ represents the total PV system output power, $N$ is the number of PV panels, and $FF$ is the fill factor computed from the maximum power point voltage $V_{\mathrm{mppt}}$, current $I_{\mathrm{mppt}}$, open-circuit voltage $V_{oc}$, and short-circuit current $I_{sc}$. The optimization function defines the dominance relations used in multi-objective optimization, where $f(\cdot )$ and $\mathrm{and} F(\cdot )$ represent objective functions, $X$ is the solution space, and $P_s^{\ast }$ denotes the set of non-dominated (Pareto-optimal) solutions.
[277]	The study assessed the approach’s efficacy during three operational periods (60, 12, and 24 min). All things considered, the Adaptive Coati Optimization algorithm presents a viable way to manage energy costs in smart homes while boosting user satisfaction and yielding financial gains.	The Adaptive Coati Optimization method was introduced.	According to the findings, tariff rates have been decreased by up to 30%, which has resulted in a 20% rise in user satisfaction and a 25% improvement in cost utilization.	$\mathrm{Initialization} (\mathrm{decision} \mathrm{variable} \mathrm{generation} Y_n:Y_{n,m}=L{B}_m+ran.\left(U{B}_m-L{B}_m\right),m=\mathrm{1,2},\dots ,j$ $\mathrm{Objective} \mathrm{function} (\mathrm{obj}) \mathrm{Minimum}{\sum }_{i=1}^I\left({\sum }_{c=1}^{Ap}{\phi }_{ap}\times Y\left(i\right)\times P_{c,i}^{cost}\right)$ $\mathrm{Definition} \mathrm{of} \mathrm{time}/\mathrm{processing} \mathrm{term} T\left(i\right)=P_{inter}\left(i\right)+P_{fix}\left(i\right)$ $\mathrm{Threshold} \mathrm{constraint} T\left(i\right)\le {\lambda }_{th}$ $\mathrm{Total} \mathrm{time} \mathrm{equality} \mathrm{constraint} T_{tot}^{Sche}=T_{tot}^{Unsche}$ $\mathrm{Total} \mathrm{energy}/\mathrm{amount} \mathrm{equality} \mathrm{constraint} A_{tot}^{Sche}=A_{tot}^{Unsche}$ $\mathrm{Index} \mathrm{bounds} \mathrm{constraint} i_{\alpha }<i<i_{\beta }$ $\mathrm{Variable} \mathrm{bounds} Y\left(i\right)\in \left[0, 1\right]$ where $Y_{n,m}$ denotes the initialized decision variable for index $n$ and dimension $m$, $L{B}_m$ and $U{B}_m$ are the lower and upper bounds, and $\mathrm{rand}\in (0, 1)$ is a uniformly distributed random number. The objective minimizes the total operational cost, where ${\varphi }_{ap}$ is a weighting coefficient and $P_{c,i}^{\cos \mathrm{t}}$ represents the cost associated with component $c$ at index $i$. $T\left(i\right)$ denotes the total processing time composed of interaction and fixed components, constrained by a threshold ${\lambda }_{th}$. The total time equality constraint and the total energy/amount equality constraint enforce consistency between scheduled and unscheduled total time and energy (or amount), respectively, while the index bounds constraint and the variable bounds ensure the feasibility of the indices and decision variables.
[278]	Optimization algorithms, including GA, CSO, and BPSO, were used to flatten energy demand profiles by considering user preferences, time considerations, and pricing signals. The effectiveness of these algorithms was compared using MATLAB simulations to identify the most effective one.	Optimization algorithms, including GA, CSO, and BPSO, were used.	GA, CSO, BPSO, VOA, and EWOA are key algorithms in reducing peak-to-average ratio (PAR) of energy consumption. VOA outperforms other algorithms without RESs, while EWOA, incorporating RESs, saved 73.8% of PAR. EWOA-based DSM costs less than non-scheduled consumption.	$\mathrm{Fitness}/\mathrm{objective} (\mathrm{energy} \cos \mathrm{t} \mathrm{over} 24 \mathrm{h}) Fitness=\sum _{i=1}^M\sum _{h=1}^{24}{E}_{ni,h}E{P}_{h } \forall h\in \left\{\mathrm{1,2},3,\dots ,24\right\}$ $\mathrm{Minimization} \mathrm{form} (\mathrm{equivalent} \cos \mathrm{t} \mathrm{objective}) min\left[\sum _{i=1}^c\sum _{h=1}^{24}{E}_{{cost}_{ni,}h}\right]$ $\mathrm{Subject} \mathrm{to}: \mathrm{Grid} \mathrm{energy} \mathrm{aggregation} \mathrm{constraint} E_{grid}=\sum _{i=1}^c\sum _{h=1}^{24}{E}_{ni,h }\forall BL$ $\mathrm{Hourly} \mathrm{energy} \mathrm{balance} (\mathrm{grid} + \mathrm{renewables} = \mathrm{total} \mathrm{load}) E_{grid,h}+E_{RES,h}=\sum _{i=1}^c\sum _{h=1}^{24}{E}_{ni,h} \forall IL$ $\mathrm{Appliance} \mathrm{ON}/\mathrm{OFF} \mathrm{state} \mathrm{constraint} \rho h,ni=\{\begin{array}{c}0\\ 1\end{array}\begin{array}{c}if appliance ni is OFF\\ if appliance ni is ON\end{array}$ where $E_{n_i,h}$ is the energy consumed by appliance $n_i$ during hour $h$ (kWh), $E{P}_h$ is the electricity price at hour $h$ (currency/kWh), $E_{n_i,h}^{\cos \mathrm{t}}=E_{n_i,h}E{P}_h$ is the operating cost of appliance $n_i$ at hour $h$ (currency), $E_{\mathrm{grid},h}$ is the grid-supplied energy at hour $h$ (kWh), $E_{\mathrm{RES},h}$ is the renewable energy supplied at hour $h$ (kWh), $E_{\mathrm{grid}}={\sum }_{h=1}^{24}{E}_{\mathrm{grid},h}$ is the total daily grid energy (kWh), $c$ is the number of appliances, $M$ is the number of scheduled devices/decision variables, and ${\rho }_{h,n_i}$ is a binary status variable indicating whether appliance $n_i$ is OFF (0) or ON (1) at hour $h$ (dimensionless).
[240]	The study explored a multi-objective hybrid optimization technique for equitable workload distribution between on peak and off-peak hours, and the concept of real-time rescheduling among home appliances using a dynamic programming strategy. It evaluated the methodology’s performance in relation to real-time pricing, time-of-use pricing, and crucial peak pricing.	A multi-objective hybrid optimization technique was used.	The proposed optimization method demonstrated relevance in reduced costs, with HAG achieving a minimum PAR of 2.22 and a cost reduction of 24.06% during scheduling, 46.14% under TOU tariff, and 29.5% in CPP cases.	$\mathrm{Candidate} \mathrm{set} \mathrm{construction} K{S}_{D_e}^{\mathrm{1,2}}=\left[T\left[\acute{i-1,\stackrel{´}{j}}\right], value \left(\acute{i-1}\right)+T\left[\acute{i-1,\stackrel{´}{ij-{List}_{time}\left(\stackrel{´}{i-1}\right)}}\right]\right]$ $\mathrm{Time} \mathrm{selection} \mathrm{rule} T\left[\acute{i,\stackrel{´}{j}}\right]=\{\begin{array}{c}min\left(K{S}_{D_e}\right) if{t}_{off}\\ min\left(K{S}_{D_e}\right) if{t}_{on}\end{array}$ $\mathrm{Binary} \mathrm{switching}/\mathrm{selection} S\left[\acute{i}\acute{j}\right]=\{\begin{array}{c}1, ifT\left[\acute{i}\acute{j}\right]==K{S}_{D_e}^2\\ 0, otherwise \end{array}$ $\mathrm{Load} \mathrm{tracking}/\mathrm{objective} \mathrm{function} min\Rightarrow F_1=\sum _{t-1}^T\left(P{L}_{sch}\left(t\right)-P{L}_{obj}\left(t\right)\right)$ $\mathrm{There} \mathrm{is} \mathrm{an} \mathrm{inverse} \mathrm{relationship} \mathrm{between} \mathrm{the} \mathrm{objective} \mathrm{load} \mathrm{power} \mathrm{and} \mathrm{electricity} \mathrm{price} \left(⋋\left(t\right)\right)which can be described mathematically as$ $\mathrm{Price}-\mathrm{objective} \mathrm{relationship} P{L}_{obj}\left(t\right)\propto {\displaystyle \frac{1}{⋋\left(t\right)}}$ $\mathrm{Scheduled} \mathrm{load} \mathrm{aggregation} P{L}_{sch}\left(t\right)=\sum _{m-1}^M{P}_{appm}\left(t\right)\times {\vartheta }_m\left(t\right)$ where $K{S}_{\left(D_e\right)}$ denotes the candidate set associated with decision element $D_e$ (dimensionless set), $T(i,j^{\prime })$ is the selected time index/value for state $(i,j^{\prime })$ (time unit consistent with the scheduling horizon, e.g., minutes or hours), $t_{\mathrm{on}}$ and $t_{\mathrm{off}}$ represent appliance ON and OFF timing states (dimensionless conditions), and $S(i^{\prime },j^{\prime })\in \{0, 1\}$ is a binary selection variable indicating whether a candidate $K{S}_{\left(D_e\right)}^2$ is chosen; $P{L}_{\mathrm{sch}}\left(t\right)$ is the scheduled load power at time $t$(kW), $P{L}_{\mathrm{obj}}(t)$ is the objective/target load power at time $t$(kW), $\lambda \left(t\right)$ is the electricity price at time $t$ (currency/kWh), $P_{\mathrm{app},m}(t)$ is the power demand of appliance $m$ at time $t$(kW), ${\vartheta }_m\left(t\right)\in \{0, 1\}$ is the ON/OFF status of appliance $m$ at time $t$ (dimensionless), and $M$ is the number of appliances.
[15]	Real-time environment and energy data were collected from embedded devices and smart meters, training a deep learning model for energy and thermal comfort prediction. The model was deployed on embedded devices for edge inference, and the whale optimization algorithm optimized occupant comfort and energy use, triggering proactive control commands using the Open Connectivity Foundation standard.	The whale optimization algorithm was used to optimize occupant comfort and energy use.	The Open Connectivity Foundation (OCF) standard was utilized for communication, and real-time OCF-based optimal actuator control tests showed effectiveness, achieving cost savings of 35.98% to 38.22%.	$\mathrm{Distance} \mathrm{vector} \mathrm{definition} \overrightarrow{D}=|\overrightarrow{C}·\overrightarrow{X^{\ast }}\left(t\right)-\overrightarrow{X}\left(t\right)|$ $\mathrm{Position} \mathrm{update} \mathrm{equation} \overrightarrow{X}\left(t+1\right)=\overrightarrow{X^{\ast }}\left(t\right)-\overrightarrow{A}·\overrightarrow{D}$ $EnergyCost function=\left(∆T\ast E{T}_o\ast T_w+∆H\ast E{H}_o\ast H_w\right)$ where $\overrightarrow{X}(t)$ and $\overrightarrow{X^{\ast }}(t)$ are the current and best solution position vectors at iteration $t$, respectively, expressed in the units of the decision variables; $\overrightarrow{D}$ is the distance vector with the same units as $\overrightarrow{X}$; $\overrightarrow{C}$ and $\overrightarrow{A}$ are coefficient vectors controlling the search behavior of the optimization algorithm; and $t$ denotes the iteration index (dimensionless). In the energy cost formulation, $\mathsf{\Delta }T$ and $\mathsf{\Delta }H$ denote the variations in temperature (°C) and humidity (%), respectively, $E{T}_o$ and $E{H}_o$ represent the energy cost coefficients associated with temperature and humidity control (e.g., kWh per unit change), and $T_w$ and $H_w$ are dimensionless weighting factors reflecting the relative importance of temperature and humidity in the total energy cost.
[32]	Smart residential homes face challenges in energy management, including efficient scheduling of electric vehicle charging and discharging, utilization of PV resources, and efficient grid power generation.	This work presented a fuzzy logic-based real-time energy management system.	The proposed energy management controller’s effectiveness was assessed on a secondary distribution system, delivering results in a mere 52 ms computational time.	$\mathrm{EV} \mathrm{charging} \mathrm{power} \mathrm{limit} 0\le P_{EV{,}_t}^{charging}\le P_{EVCHG,}$ $\mathrm{EV} \mathrm{discharging} \mathrm{power} \mathrm{limit} 0\le P_{EV{,}_t}^{decharging}\le P_{EVCHG,}$ $\mathrm{SOC} \mathrm{bounds} SO{C}_{min}\le RSO{C}_i\le SO{C}_{max,}$ $\mathrm{SOC} \mathrm{update} \mathrm{equation} RSO{C}_t=RSO{C}_{t-i}-{\displaystyle \frac{P_{EV{,}_t}\times ∆t}{EV{C}_{ap}\times T}},$ $\mathrm{Charging} \mathrm{time} \mathrm{remaining} (\mathrm{CTR}) CT{R}_t={\displaystyle \frac{EV{C}_{ap}\left(1-RSO{C}_t\right)}{P_{EVCHG}}}$ Transformer (DT) loading indicator rules $Normal DT Load indicator=\{\begin{array}{c}1\\ 0\end{array}{\displaystyle \genfrac{}{}{0pt}{}{ DT Loading\left(p.u\right)\le 1.}{ DT Loading\left(p.u\right)\ge 1.}}$ $DT Overload indicator=\{\begin{array}{c}0\\ 1\end{array}{\displaystyle \genfrac{}{}{0pt}{}{ DT Loading\left(p.u\right)\le 1.}{ DT Loading\left(p.u\right)\ge 1.}}$ Incentive membership functions (fuzzy) $Low Incentive={\displaystyle \frac{20-Incentive}{20}}, 0\le Incentive\left(cents\right)\le 20$ $High Incentive={\displaystyle \frac{Incentive}{20}}, 0\le Incentive\left(cents\right)\le 20.$ Charging membership functions (fuzzy) $Low Charging={\displaystyle \frac{0.25-EVCM}{0.25}}, 0\le EVCM\le 0.25.$ $Medium Charging=\{\begin{array}{c}{\displaystyle \frac{EVCM-0.1}{0.4}}, 0.1\le EVCM\le 0.5.\\ {\displaystyle \frac{0.9-EVCM}{0.4}}, 0.5\le EVCM\le 0.9.\end{array}$ $High Charging={\displaystyle \frac{EVCM-0.75}{0.25}}, 0.75\le EVCM\le 1$ where $P_{EV,t}^{\mathrm{ch}}$ and $P_{EV,t}^{\mathrm{dis}}$ denote EV charging and discharging power at time $t$(kW), $P_{EVCHG}$ is the maximum EV charging/discharging power (kW), $RSO{C}_t$ is the EV state of charge at time $t$ (dimensionless, 0–1), $SO{C}_{\mathrm{m}\mathrm{i}\mathrm{n}}$ and $SO{C}_{\mathrm{m}\mathrm{a}\mathrm{x}}$ are SOC limits (dimensionless), $EV{C}_{ap}$ is the EV battery capacity (kWh), $\mathsf{\Delta }t$ is the time-step duration (h), $CT{R}_t$ is the remaining charging time (h), $D{T}_{\mathrm{Loading}} \left(\mathrm{p}.\mathrm{u}.\right)$ is the distribution transformer loading in per-unit (dimensionless), Incentive is the incentive level (cents), and $EVCM$ is a normalized charging measure (dimensionless, 0–1); $\mu (\cdot )$ denotes fuzzy membership degree (dimensionless).
[279]	The Intelligent Smart Energy Management System (ISEMS) aims to accurately estimate energy availability and plan for the future in a smart grid environment incorporating renewable energy.	The Support Vector Machine regression model based on PSO showed superior performance accuracy.	The experimental setup for ISEMS was demonstrated, and evaluated in various configurations, and IoT integration was implemented for user comfort monitoring.	$\mathrm{Mean} \mathrm{Absolute} \mathrm{Error} (\mathrm{M}\mathrm{A}\mathrm{E})={\displaystyle \frac{1}{N}}\sum _{i=1}^N\left|\overrightarrow{Pi}-Pi\right|$ $\mathrm{Mean} \mathrm{Absolute} \mathrm{Percentage} \mathrm{Error} (MAPE)={\displaystyle \frac{1}{N}}\sum _{i=1}^{N}100{\displaystyle \frac{\left|\overrightarrow{Pi}-Pi\right|}{Pn}}$ $\mathrm{Root} \mathrm{Mean} \mathrm{Square} \mathrm{Error} (RMSE)=\sqrt{{\displaystyle \frac{1}{N}}\sum _{i=1}^N\left|\overrightarrow{Pi}-Pi\right|}$ where $N$ is the number of samples (dimensionless), $P_i$ is the measured value in its physical unit (e.g., kW or kWh), and ${\widehat{P}}_i$ is the corresponding predicted value in the same unit, while MAE and RMSE retain the unit of $P$ and MAPE is reported as a percentage.
[280]	The management method was divided into two phases based on charging constraints and initial charge status. Stage A includes three operational states based on PV generation availability, while Stage B predicts five operational states. This advanced control strategy ensures precise energy absorption by EVs, minimizes residential electricity costs, and synchronizes electrical load profiles.	The system integrated EVs and optimizing residential electricity expenses using TOU cost, usage fluctuations, PV generation patterns, and EV variables.	The proposed plan shows a significant reduction in residential electricity expenses and normalization of power load characteristics, with the proposed plan for intelligent households using both EVs and PV generation outperforming EV-only households.	Stage A: Load EV interaction $\mathrm{Mode} 1 (\mathrm{Baseline} \mathrm{load}) P^{L\text{-}New}\left(t\right)=P^L\left(t\right)$ $\mathrm{Mode} 2 (\mathrm{Grid}–\mathrm{EV} \mathrm{support}) P^{GEV}\left(t\right)=P^{Avg}-P^L\left(t\right)$ $P^{L\text{-}New}\left(t\right)=P^L\left(t\right)+P^{GEV}\left(t\right)$ $\mathrm{Mode} 3 (\mathrm{EV} \mathrm{load} \mathrm{shifting}) P^{EVL}\left(t\right)=P^L\left(t\right)-P^{Avg}$ $P^{L\text{-}New}\left(t\right)=P^L\left(t\right)+P^{EVL}\left(t\right)$ Stage B: PV Load EV interaction Mode 4 (PV to load) $P^{PVL}\left(t\right)=P^{PV}\left(t\right)$ $P^{L\text{-}New}\left(t\right)=P^L\left(t\right)+P^{PVL}\left(t\right)$ Mode 5 (PV to EV) $P^{PVEV}\left(t\right)=P^{PV}\left(t\right)$ Mode 6 (PV + Grid–EV coordination) $P^{L\text{-}New}\left(t\right)=P^L\left(t\right)+P^{GEV}\left(t\right)$ $P^{PVEV}\left(t\right)=P^{PV}\left(t\right)$ $P^{GEV}\left(t\right)=P^{Avg}-P^L\left(t\right)-P^{PVEV}\left(t\right)$ Mode7 (PV-assisted load balancing) $P^{PVL}\left(t\right)=P^L\left(t\right)-P^{Avg}$ $P^{PVEV}\left(t\right)=P^{PV}\left(t\right)-P^{PVL}\left(t\right)$ $P^{L\text{-}New}\left(t\right)=P^L\left(t\right)+P^{PVL}\left(t\right)$ Mode 8 (High PV penetration) $P^{PVL}\left(t\right)=P^{PV}\left(t\right)$ $P^{PVEV}\left(t\right)=P^{PV}\left(t\right)-P^{Avg}-P^{PVL}\left(t\right)$ $P^{L\text{-}New}\left(t\right)=P^L\left(t\right)-2\times P^{EVL}\left(t\right)$ Total daily cost function $C_{III}^{Total}=P{V}^{C-Day}\left(t\right){\displaystyle \sum _{t=1}^{24}}\left[C\left(t\right)\times \left(P^L\left(t\right)+P^{GEV}\left(t\right)-P^{EVL}\left(t\right)-P^{PVL}\left(t\right)-P^{PVEV}\left(t\right)\right)\right]$ where ${\mathrm{P}}^{\mathrm{L}}(\mathrm{t})$ is the baseline household load at time $\mathrm{t}$(kW), ${\mathrm{P}}^{\mathrm{L}\text{-}\mathrm{New}}(\mathrm{t})$ is the updated load after coordination (kW), ${\mathrm{P}}^{\mathrm{A}\mathrm{v}\mathrm{g}}$ is the average load over the scheduling horizon (kW), ${\mathrm{P}}^{\mathrm{G}\mathrm{E}\mathrm{V}}(\mathrm{t})$ denotes grid-to-EV power exchange (kW), ${\mathrm{P}}^{\mathrm{E}\mathrm{V}\mathrm{L}}(\mathrm{t})$ represents EV-induced load variation (kW), ${\mathrm{P}}^{\mathrm{P}\mathrm{V}}(\mathrm{t})$ is photovoltaic generation power (kW), ${\mathrm{P}}^{\mathrm{P}\mathrm{V}\mathrm{L}}(\mathrm{t})$ is PV power supplied to the household load (kW), ${\mathrm{P}}^{\mathrm{P}\mathrm{V}\mathrm{E}\mathrm{V}}(\mathrm{t})$ is PV power supplied to EV charging (kW), $\mathrm{C}\left(\mathrm{t}\right)$ is the electricity price at time $\mathrm{t}$ (currency/kWh), and $\mathrm{t}$ denotes the hourly time index over a 24 h horizon.
[281]	This paper used historical weather data to predict PV power outcomes, analyzing the benefits of HEMSs with PV and battery ESSs for peak load shaving and grid stability.	This paper proposed an intelligent HEMS with three adjustable strategies to maximize economic benefits and consumer comfort. It introduces a novel objective function focusing on satisfying users’ needs, integrating a tri-objective model into an algorithm.	The proposed model significantly reduced electricity bill expenditure by 39.81% and maintained grid balance by reducing peak load by 50.37%, resulting in a 1.6-fold improvement in user comfort index.	$\mathrm{Appliance} \mathrm{ON}/\mathrm{OFF} \mathrm{indicator} X_{i,j}\{\begin{array}{c}0,\\ 1,\end{array}{\displaystyle \genfrac{}{}{0pt}{}{ j\notin \left[t_{i,}{t}_i+L_i\right]}{ j\in \left[t_{i,}{t}_i+L_i\right]}}$ $\mathrm{Daily} \mathrm{schedule} \mathrm{matrix} X_{day}^{total}=\left[\begin{array}{ccc}{X}_{\mathrm{1,1}}\dots & \dots \dots & X_{\mathrm{1,48}}\\ X_{i,j}\dots & X_{i,j}\dots & X_{i,48}\\ X_{n,1}\dots & \dots \cdots & X_{n,48}\end{array}\right]$ $\mathrm{Time} \mathrm{index} \mathrm{constraints} 1\le a_i\le b_i\le 48,t_i\in \left[a_i,b_i\right]$ photovoltaics $\mathrm{PV} \mathrm{output} \mathrm{power} P{V}_j=\eta S_{array}{I}_j\left(1-0.005\left(T_j+25\right)\right)$ Battery $\mathrm{Battery} \mathrm{state} \mathrm{of} \mathrm{charge} \mathrm{update} So{C}_j=K_d·So{C}_{j-1}+Charg{e}_j-{Discharge}_j$ Tariff/objective index $Ta{r}_{max}={\gamma }_1{\displaystyle \frac{Sat}{{Sat}_{max}}}-{\gamma }_2{\displaystyle \frac{{\sigma }_m}{{\sigma }_{m,max}}}-{\gamma }_3{\displaystyle \frac{N}{N_{max}}}$ where $X_{i,j}\in \{0, 1\}$ indicates whether appliance $i$ is ON in time slot $j$ (dimensionless), $L_i$ is the operating duration in slots (dimensionless), $t_i$ is the start slot index (dimensionless), $a_i$ and $b_i$ define the allowable start-time window (dimensionless), $n$ is the number of appliances (dimensionless), and the horizon is divided into 48 slots (e.g., 30 min intervals); $P{V}_j$ is PV output power at slot $j$ (kW), $\eta$ is PV efficiency (dimensionless), $S_{\mathrm{array}}$ is PV array area (m2), $I_j$ is solar irradiance (kW/m2 or W/m2, use one consistently), and $T_j$ is PV cell/ambient temperature (°C); $So{C}_j$ is battery state of charge at slot $j$ (dimensionless or kWh depending on definition), $K_d$ is the self-discharge/retention factor (dimensionless), and $Charg{e}_j$ and $Discharg{e}_j$ are charged and discharged energy per slot (kWh); $Ta{r}_{\mathrm{m}\mathrm{a}\mathrm{x}}$ is a dimensionless tariff index, ${\gamma }_1,{\gamma }_2,{\gamma }_3$ are dimensionless weights, and $Sat$, ${\sigma }_m$, and $N$ are normalized performance measures scaled by their corresponding maxima.
[282]	The study employed a data analytics platform utilizing ANN and a PSO algorithm to collect real-time ambient data and control air conditioner operation, predicting power consumption, indoor temperature, and humidity accurately.	A PSO algorithm was used.	The intelligent cooling management system, tested in a smart home environment using an 8000 BTU air conditioner, predicted air conditioner behavior and ambient data, indicating potential energy savings in smart home applications, with validation results proving its effectiveness.	$\mathrm{Thermal} \mathrm{comfort} \mathrm{penalty} \mathrm{function} penalt{y}_i\left(t\right)=\{\begin{array}{c}100 \\ 0\end{array}{\displaystyle \genfrac{}{}{0pt}{}{\times \left|PM{V}_i\left(t\right)\right| ,-0.5<PM{V}_i}{,PM{V}_i\left(t\right)<-0.5 or PM{V}_i}}$ ${\displaystyle \genfrac{}{}{0pt}{}{<0.5}{>0.5}}$; $i=\mathrm{1,2},\dots ,n,t=\mathrm{1,2},\dots ,k$ $\mathrm{Penalized} \mathrm{fitness} \mathrm{value} P_i\left(t\right)=p_i^{avg}\left(t\right)+penalt{y}_i\left(t\right) ;i=\mathrm{1,2},\dots ,n,t=\mathrm{1,2},\dots ,k$ $\mathrm{Personal} \mathrm{best} \mathrm{fitness} P{best}_i=min\left[P_i\left(t\right)\right] ;i=\mathrm{1,2},\dots ,n,t=\mathrm{1,2}\dots ,k$ $\mathrm{Global} \mathrm{best} \mathrm{fitness} Gbest\left(t\right)=min\left[{Pbest}_i\left(t\right)\right] ;i=\mathrm{1,2},\dots ,n,t=\mathrm{1,2}\dots ,k$ Particle swarm optimization (PSO) update rules $\mathrm{Velocity} \mathrm{update}: {\overrightarrow{V}}_i\left(t+1\right)=w{\overrightarrow{V}}_i\left(t\right)+n_i{r}_i\left[P{best}_i-{\overrightarrow{X}}_i\left(t\right)\right]+n_2{r}_2\left[G{best}_i-{\overrightarrow{X}}_i\left(t\right)\right] ;$ $i=\mathrm{1,2},\dots ,n,t=\mathrm{1,2}\dots ,k$ Position update: ${\overrightarrow{X}}_i\left(t+1\right)={\overrightarrow{X}}_i\left(t+1\right)+{\overrightarrow{V}}_i\left(t+1\right) ;$ $i=\mathrm{1,2},\dots ,n,t=\mathrm{1,2}\dots ,k$ where $PM{V}_i\left(t\right)$ is the predicted mean vote comfort index of particle $i$ at time $t$ (dimensionless), $penalt{y}_i(t)$ is a comfort penalty term (same unit as the fitness value), $p_i^{avg}(t)$ is the average objective cost of particle $i$ at time $t$ (e.g., currency), $P_i\left(t\right)$ is the penalized fitness value (currency), $Pbes{t}_i$ is the personal best fitness of particle $i$, and $Gbest$ is the global best fitness among all particles; ${\overrightarrow{X}}_i(t)$ and ${\overrightarrow{V}}_i(t)$ are the position and velocity vectors of particle $i$ at iteration $t$ (with units matching the decision variables), $w$ is the inertia weight (dimensionless), $n_1$ and $n_2$ are cognitive and social acceleration coefficients (dimensionless), $r_1$ and $r_2$ are random numbers uniformly distributed in $[0,1]$ (dimensionless), $n$ is the number of particles, and $k$ is the maximum number of iterations.
[283]	Demand-side management (DSM) is a crucial aspect of microgrid and smart grid technology, aimed at controlling requirements while maintaining client trust. The research focused on helping households manage their power plans.	The HBA + DMO technique, combining Honey Badger Optimization (HBA) and Dwarf Mongoose Optimization (DMO), was used.	The proposed approach collects energy data for reporting, monitoring, and engagement, with a computational time of approximately 213.42. This can help reduce waiting times and improve user comfort in various settings.	$\mathrm{Objective} \mathrm{function} Minimize\sum _{n=1}^N\left(\sum _{s=1}^{App}{\varnothing }_{App}\times Z\left(n\right)\times F_{s,n}^{\cos t}\right)$ $\mathrm{Definition} \mathrm{of} I\left(n\right)=F_{sta}\left(n\right)+F_{move}\left(n\right)+F_{fix}\left(n\right)$ $\mathrm{Threshold} \mathrm{constraint} I\left(n\right)\le {\lambda }_{th}$ $\mathrm{Total} P$ $\mathrm{constraint} I_{total}^{Sche}\le I_{total}^{Unsche}$ Index bounds $P_{total}^{Sche}\le P_{total}^{Unsche}$ Decision-variable bound $n\alpha <n<n_{\beta }$ $Z\left(n\right)\in \left[0, 1\right]$ where $Z\left(n\right)$ is a bounded decision variable (dimensionless), ${\varphi }_{App}$ is a weighting coefficient (dimensionless), $F_{s,n}^{cost}$ is the cost associated with appliance/state $s$ sat index $n$ (currency), $I\left(n\right)$ is an aggregate index formed by stationary moving and fixed components $F_{sta}\left(n\right)$, $F_{move}\left(n\right)$, and $F_{fix}\left(n\right)$ (same unit as $I$), ${\lambda }_{th}$ is the corresponding threshold (same unit as $I$), $I_{total}^{Sche}$ and $I_{total}^{Unsche}$ denote the total scheduled and unscheduled values of $I$, $P_{total}^{Sche}$ and $P_{total}^{Unsche}$ denote the total scheduled and unscheduled power/energy metric (kW or kWh as defined in the study), $N$ is the total number of indices, $App$ is the number of appliances/states, and $n$ is the index variable.
[284]	The model integrated renewable energy, PV systems, wind power, and an energy storage system to ensured coordinated electricity flow in residential houses. It used demand response schemes and a dynamic model for the System Performance Index. It also introduced a Dynamic Distributed Energy Storage Strategy and a Wild Mice Colony optimization algorithm.	A Wild Mice Colony optimization algorithm was used.	The strategy of DDESS could significantly reduced energy consumption by over 100% of load demand, optimize the energy system, and ensure synchronization, thereby minimizing EC costs from the PG.	$\mathrm{Demand} \mathrm{side} \mathrm{energy} \mathrm{balance} (\mathrm{general} \mathrm{form}) E_d\left(t\right)=E_{wt}\left(t\right)+E_{pv}\left(t\right)+E_{gr}\left(t\right)+E_{dc}\left(t\right)$ $\mathrm{Operating} \mathrm{supply} \mathrm{definition} E_d\left(t\right)=E_{wt}\left(t\right)+E_{pv}\left(t\right)+E_{dc}\left(t\right)$ $\mathrm{Utility}\text{-}\mathrm{grid} \mathrm{exchange} (\mathrm{case} \mathrm{rule}) E_{op}\left(t\right)=E_{wt}\left(t\right)+E_{pv}\left(t\right)+E_{dc}\left(t\right)$ $E_{ug}\left(t\right)=E_{gr}\left(t\right) if E_{op}\left(t\right)=0$ $E_{ug}\left(t\right)={-E}_{op}\left(t\right) if E_{gr}\left(t\right)=0$ Cost components $C_{gr}\left(t\right)=P_{r_{gr}}{\displaystyle \sum _{t=1}^N}{E}_{gr}\left(t\right)$ $C_{wt}\left(t\right)=P_{r_{DER}}{\displaystyle \sum _{t=1}^N}{E}_{pv}\left(t\right)$ $C_b\left(t\right)=P_{r_{DER}}{\displaystyle \sum _{t=1}^N}{E}_b\left(t\right)$ $C_{op}\left(t\right)=P_{r_{DER}}{\displaystyle \sum _{t=1}^N}{E}_{op}\left(t\right)$ Optimization objective function $\min J\left(K\right)={\displaystyle \sum _{k=1}^N}{\displaystyle \sum _{j=1}^{N_c=k}}\left[C_{ug,kj}\left(k\right)-C_{wt,kj}\left(k\right)-C_{pv,kj}\left(k\right)-C_{b,kj}\left(k\right)-C_{op,kj}\left(k\right)\right]$ $\min J\left(K\right)={\displaystyle \sum _{k=1}^N}{\displaystyle \sum _{j=1}^{N_c=k}}\left[P_{r_{gr}}\left(E_{ug,kj}\left(k\right)-E_{op,kj}\left(k\right)\right)-P_{r_{DER}}{E}_{DER,kj}\left(k\right)\right]$ where $E_d\left(t\right)$ is the total demand energy at time $t$ (kWh), $E_{wt}\left(t\right)$, $E_{pv}(t)$, $E_{gr}(t)$, and $E_{dc}\left(t\right)$ denote energy contributions from wind, photovoltaic, grid import, and dispatchable/source component, respectively (kWh), $E_{op}(t)$ is the total local operating energy supply (kWh), and $E_{ug}(t)$ is the net utility-grid exchange (kWh, positive for import and negative for export); $C_{gr}$, $C_{pv}$, $C_b$, and $C_{op}$ are the corresponding cost terms (currency), $P_{r\_gr}$ and $P_{r\_DER}$ are electricity price rates for grid and DER energy (currency/kWh), $N$ is the number of time steps (dimensionless), $k$ and $j$ are index variables (dimensionless), and $N_c\left(k\right)$ is the number of considered components/cases at step $k$
[223]	Smart home users face high monthly energy consumption bills and struggle to optimize devices, increasing energy demand and accelerating global greenhouse effects. Uneven usage of non-shiftable appliances can exceed power limits, leading to short blackouts. To address these challenges, a mobile application has been developed to effectively control energy consumption in smart homes.	The PSO algorithm was used.	Two tests were carried out: one with and one without the PSO algorithm being used. The outcome demonstrated that the PSO algorithm outperforms the others in terms of energy consumption optimization.	$\mathrm{Ackley} \mathrm{benchmark} \mathrm{function} f\left(x\right)=-aexp\left(-b\sqrt{{\displaystyle \frac{1}{d}}\sum _{i=1}^d{x}_i^2}\right)-exp\left({\displaystyle \frac{1}{d}}\sum _{i=1}^d\cos \left({CS}_i\right)\right)+a+exp$ It is a common practice to test optimization problems using the Ackley equation [11]. Based on the author, the recommended unknown variable values inside the Ackley equation are a = 20; b = 0.2; c = 2π. $\mathrm{where} \mathrm{x}=(x_1,x_2,\dots ,x_d)$ is a $d$-dimensional decision vector, $x_i$ denotes the $i$-th decision variable, $d$ is the problem dimensionality, and $a$, $b$, and $c$ are predefined constants controlling the shape of the function; following standard practice in optimization benchmarking, the recommended parameter values are $a=20$, $b=0.2$, and $c=2\pi$, yielding a multimodal landscape commonly used to evaluate global optimization performance.
[285]	This paper proposes a flexible smart home energy management framework, considering various technologies like CHP units, PV generation units, and electric vehicles, to optimize energy payment and user satisfaction.	The proposed model uses a multi-criteria decision making (MCDM) approach and a mixed-integer linear programming (MILP) problem, solved by the CPLEX solver in the GAMS environment.	The study revealed significant disparities in energy payments when the flexibility limit falls below 40%, affecting end-user satisfaction and self-sufficiency, with variations of 27.4%, 100%, and 56.64%, respectively.	$\mathrm{Objective} \mathrm{function} f_1=Min\sum _{t=1}^{N_t}\sum _{s=1}^{N_s}{\rho }_s\left(C_{t,s}^E+C_{t,s}^T+C_{t,s}^G\right)$ Where $\mathrm{Electricity} \mathrm{trading} \mathrm{cost} C_{t,s}^E=\left({Pe}_t^{buy,grid}{P}_{t,s}^{buy,grid}-{Pe}_t^{sell,grid}{P}_{t,s}^{sell,grid}\right)+\left({Pe}_t^{buy,EC}{P}_{t,s}^{buy,EC}-{Pe}_t^{sell,EC}{P}_{t,s}^{sell,EC}\right)$ Thermal energy cost $C_{t,s}^T={Ph}_t^{buy,grid}{H}_{t,s}^{buy,grid}-{Ph}_t^{sell,grid}{H}_{t,s}^{sell,grid}$ $\mathrm{Gas} \mathrm{energy} \mathrm{cost} C_{t,s}^G={Pg}^{buy,grid}{G}_{t,s}^{buy,grid}$ $\mathrm{where} t$ $\mathrm{and} s$ $\mathrm{denote} \mathrm{the} \mathrm{time} \mathrm{step} \mathrm{and} \mathrm{scenario} \mathrm{index} (\mathrm{dimensionless}), {\rho }_s$ $\mathrm{is} \mathrm{the} \mathrm{probability} \mathrm{of} \mathrm{scenario} s$, $P_{t,s}$, $H_{t,s}$, $\mathrm{and} G_{t,s}$ $\mathrm{represent} \mathrm{electricity}, \mathrm{thermal} \mathrm{energy}, \mathrm{and} \mathrm{gas} \mathrm{quantities} \mathrm{traded} \mathrm{at} \mathrm{time} t$ $\mathrm{under} \mathrm{scenario} s$ $\left(\mathrm{kWh}\right), P{e}_t$, $P{h}_t$, $\mathrm{and} Pg$ $\mathrm{are} \mathrm{the} \mathrm{corresponding} \mathrm{electricity}, \mathrm{heat}, \mathrm{and} \mathrm{gas} \mathrm{prices} (\mathrm{currency}/\mathrm{kWh}), \mathrm{and} C_{t,s}^E$, $C_{t,s}^T$, $\mathrm{and} C_{t,s}^G$ denote the electricity, thermal, and gas cost components (currency).
[15]	Real-time environment and energy data were collected from embedded devices and smart meters, training a deep learning model for energy and thermal comfort prediction. The model was deployed on embedded devices for edge inference, with the WOA optimizing occupant comfort and energy use.	The whale optimization algorithm was introduced.	The whale optimization algorithm generates results for a fuzzy logic controller, which activates control commands for proactive response. Real-time OCF-based experiments confirm system efficacy, achieving cost savings of 35.98% to 38.22%.	Optimization update equations $\mathrm{Distance} \mathrm{vector} \overrightarrow{D}=\left|\overrightarrow{C}\cdot \overrightarrow{x^{\ast }}\left(t\right)-\overrightarrow{x}\left(t\right)\right|$ $\mathrm{Position} \mathrm{update} \overrightarrow{x}\left(t+1\right)=\overrightarrow{x^{\ast }-}\overrightarrow{A}·\overrightarrow{D}$ $EnergyCost=\left(∆T\ast E{T}_0\ast T_w+∆H\ast E{H}_0\ast H_w\right)\ast Price$ $\mathrm{Humidity} \mathrm{deviation} \Delta H=\{{\displaystyle \genfrac{}{}{0pt}{}{|H_c-P_{H_{min}}| if H_c<P_{H_{min}}}{|H_c-P_{H_{max}}| if H_c>P_{H_{min}}}}$ $\mathrm{Temperature} \mathrm{deviation} \Delta T=\{{\displaystyle \genfrac{}{}{0pt}{}{|T_c-P_{T_{min}}| if T_c<P_{T_{min}}}{|T_c-P_{T_{max}}| if T_c>P_{T_{min}}}}$ where $\overrightarrow{x}(t)$ and $\overrightarrow{x^{\ast }}(t)$ denote the current and best solution vectors at iteration $t$, respectively, $\overrightarrow{D}$ is the distance vector with units consistent with the decision variables, and $\overrightarrow{A}$ and $\overrightarrow{C}$ are dimensionless coefficient vectors controlling the optimization process; $\mathsf{\Delta }T$ and $\mathsf{\Delta }H$ represent deviations of indoor temperature (°C) and humidity (%) from their comfort bounds $[T_{min},T_{max}]$ and $[H_{min},H_{max}]$, $E{T}_0$ and $E{H}_0$ are energy conversion coefficients (kWh per unit deviation), $T_w$ and $H_w$ are dimensionless weighting factors, and Price denotes the electricity tariff (currency/kWh).
[286]	This paper investigates the integration of attention networks in home energy management systems (HEMSs) to improve energy consumption optimization. It examines the AMpds2 dataset and compares its performance across various forecasting methodologies, using metrics like RMSE and MAE, and advanced optimizers.	The proposed solution uses attention networks to dynamically allocate energy consumption significance, focusing on the AMpds2 dataset and assesses performance across various time series forecasting methodologies.	The study analyzed 16 hyperparameter combinations across four-time series models and found that transformers improved energy and load pattern forecasting accuracy by 4%, using Python 3.2 and the matplotlib library.	$\mathrm{Gradient} \mathrm{evaluation} H\left(t\right)=f^{\prime }\left(y\left(t-1\right)\right)$ $\mathrm{First}\text{-}\mathrm{moment} \mathrm{estimate} n\left(t\right)={\delta }_1\ast m\left(t-1\right)+\left(1-{\delta }_1\right)\ast H\left(t\right)$ $\mathrm{Second}\text{-}\mathrm{moment} \mathrm{estimate} v\left(t\right)=max({\delta }_2\ast u\left(t-1\right),abs\left(H\left(t\right))\right)$ $\mathrm{Bias}\text{-}\mathrm{corrected} \mathrm{scaling} \mathrm{term} n\left(t\right)={\displaystyle \frac{\gamma }{1-{\delta }_1\left(t\right)}}$ The gradient used to update the parameter is computed in the following manner. $\mathsf{\Phi }\left(t\right)={\displaystyle \frac{n\left(t\right)}{v\left(t\right)}}$ $y\left(t\right)= y\left(t-1\right)-\eta \left(t\right)\ast \Phi \left(t\right)$ Alternatively, the comprehensive update equation is expressed as: $y(t)= y(t-1)-{\displaystyle \frac{\gamma }{1-{\delta }_1\left(t\right)}}\ast {\displaystyle \frac{n\left(t\right)}{v\left(t\right)}}$ Attention mechanism (linear projections) $X\mathrm{1,2},\dots .n\ast Wq=Q\mathrm{1,2}\dots .,n$ $.X\mathrm{1,2},\dots .n\ast Wk=K\mathrm{1,2}\dots .,n$ $.X\mathrm{1,2},\dots .n\ast Wv =V\mathrm{1,2}\dots .,n.$ Step 2 $Q1\ast K\mathrm{1,2}\dots .,n=S1$ $.Q2\ast K\mathrm{1,2}\dots .,n=S2$ $\dots Qn\ast K\mathrm{1,2}\dots .,n = Sn.$ where $y\left(t\right)$ denotes the model parameter at iteration $t$, $f^{\prime }(\cdot )$ is the gradient of the objective function, $H\left(t\right)$ is the instantaneous gradient, $n\left(t\right)$ and $v\left(t\right)$ are first- and second-moment estimates, ${\delta }_1$ and ${\delta }_2$ are decay coefficients (dimensionless), $\gamma$ is a scaling factor, and $\eta \left(t\right)$ is the learning rate; $\mathrm{X}$ is the input feature matrix, ${\mathrm{W}}_q$, ${\mathrm{W}}_k$, and ${\mathrm{W}}_v$ are learnable projection matrices, $\mathrm{Q}$, $\mathrm{K}$, and $\mathrm{V}$ are the query, key, and value matrices, and ${\mathrm{S}}_i$ denotes the similarity score for the $i$-th query.
[287]	The paper proposed a flexible approach for aggregators in distribution systems, utilizing load flexibility resources to enable real-time appliance rescheduling and shifting to meet demands.	A new Reinforced Learning Quantum Inspired Grey Wolf Optimization (RLQIGWO) was used.	RLQIGWO, a grey wolf optimizer, integrated reinforcement learning and quantum mechanics principles achieved better performance in load balancing, resource utilization, and task execution, enhancing energy management strategies.	Wolves encircle prey using the following equations: $D=\left|C\cdot X_p-X\right|$ $X\left(t+1\right)=X_p-A·D$ $X_1=X_{\alpha }- A_1\cdot |C_1\cdot X_{\alpha } X|$ $X_2=X_{\beta }- A_2\cdot |C_2\cdot X_{\alpha } X|$ $X_3=X_{\alpha }- A_3\cdot \left|C_3\cdot X_{\alpha } X\right|$ $X\left(t+1\right)={\displaystyle \frac{X_1+X_2+X_3}{3}}$ Reinforced-Learning-based position update $Q\left(s,a\right)\leftarrow Q\left(s,a\right)+\alpha \left[r+\gamma max{a}^{\prime }Q\left(s^{\prime },a^{\prime }\right)Q\left(s,a\right)\right]$ $Q\left(i,j\right)\leftarrow Q\left(i,j\right)+\alpha \left(f\left(i\right)-Q\left(i,j\right)\right)$ Quantum-inspired based position update $\left|\Psi ⟩=\alpha |0⟩+\beta \right|1⟩$ $\mathrm{Where} {\left|\alpha \right|}^2+{\left|\beta \right|}^2$ $\theta =2\pi ·rand$ $r=\left|sin\left(\theta \right)·X_1+\cos \left(\theta \right)·X_2+X_3\right|$ $x^{\prime }\left(i,j\right)=r·\left(ub\left(j\right)-lb\left(j\right)\right)+lb\left(j\right)$ $PrefA\left(i\right)= \left[t_{allow} \left(i\right), D_{allow}\left(i\right), C_A \left(i\right)\right]$ $t_{allow} \left(i\right)\le t_{new} \left(i\right)\le t_{allow} \left(i\right)D_{allow} \left(i\right)$ where $\mathrm{X}\left(t\right)$ and $\mathrm{X}(t+1)$ denote the current and updated position vectors of a search agent, ${\mathrm{X}}_p$ represents the prey (best solution), ${\mathrm{X}}_{\alpha }$, ${\mathrm{X}}_{\beta }$, and ${\mathrm{X}}_{\delta }$ are the leading wolf positions, and $\mathrm{A}$ and $\mathrm{C}$ are coefficient vectors controlling exploration and exploitation (dimensionless); $Q(s,a)$ denotes the action–value function in reinforcement learning, $\alpha$ is the learning rate, $\gamma$ is the discount factor, and $r$ is the received reward; $\mid \mathsf{\Psi }⟩$ is a quantum state with probability amplitudes $\alpha$ and $\beta$, $\theta$ is a random rotation angle, $lb\left(j\right)$ and $ub\left(j\right)$ are the lower and upper bounds of the $j$-th decision variable, and all position vectors are expressed in units consistent with the optimization variables.
[219]	The paper proposed a demand response method for managing residential energy consumption, aiming to reduce costs, the peak-to-average ratio, and imports, addressing the growing energy consumption in the residential sector.	Manta ray foraging optimization (MRFO) and long-term memory MRFO (LMMRFO) algorithms were used.	The proposed plan effectively reduced electricity costs and maximized profit through case studies and comparative studies, demonstrating the legality and effectiveness of LMMRFO and MRFO.	Manta Ray foraging optimization Chain harvesting: $x_i^d\left(t+1\right)=\{\begin{array}{c}{x}_i^d\\ x_i^d\end{array}{\displaystyle \genfrac{}{}{0pt}{}{\left(t\right)+r·\left(x_{best}^d\left(t\right)\right)+\alpha ·}{\left(t\right)+r·\left(x_{i-1}^d\left(t\right)\right)+\alpha ·}}$ ${\displaystyle \genfrac{}{}{0pt}{}{\left(x_{best}^d\left(t\right)-x_i^d\left(t\right)\right) i=1}{\left(x_{best}^d\left(t\right)-x_i^d\left(t\right)\right) i=2,\dots N}}$ $\mathsf{\alpha }=2.\mathrm{r}\sqrt{\left|\log \left(r\right)\right|}$ Cyclone foraging $\{\begin{array}{c}\begin{array}{c}{X}_i\left(t+1\right)=X_{best}+r·\left(X_{i-1}\left(t\right)-X_i\left(t\right)\right)\\ Y_i\left(t+1\right)=Y_{best}+r·\left(Y_{i-1}\left(t\right)-Y_i\left(t\right)\right)\end{array}\\ {\displaystyle \genfrac{}{}{0pt}{}{+e^{bw}·\cos \left(2\pi w\right)·\left(X_{best}-X_i\left(t\right)\right)}{+e^{bw}·\sin \left(2\pi w\right)·\left(Y_{best}-Y_i\left(t\right)\right)}}\end{array}$ $x_i^d\left(t+1\right)=\{\begin{array}{c}{x}_{best}^d\\ x_{best}^d\end{array}{\displaystyle \genfrac{}{}{0pt}{}{\left(t\right)+r·\left(x_{best}^d\left(t\right)-x_{i-1}^d\left(t\right)\right)}{\left(t\right)+r·\left(x_{i-1}^d\left(t\right)-x_{i-1}^d\left(t\right)\right)}}$ ${\displaystyle \genfrac{}{}{0pt}{}{+\beta ·}{+\beta ·}}{\displaystyle \genfrac{}{}{0pt}{}{\left(x_{best}^d\left(t\right)-x_i^d\left(t\right)\right) i=1}{\left(x_{best}^d\left(t\right)-x_i^d\left(t\right)\right) i=2,\dots N}}$ $x_{rand}^d={Lb}^d+r·\left({Ub}^d-L{b}^d\right)$ $x_i^d\left(t+1\right)=\{\begin{array}{c}{x}_{rand}^d\\ x_{rand}^d\end{array}{\displaystyle \genfrac{}{}{0pt}{}{\left(t\right)+r·\left(x_{rand}^d\left(t\right)+x_i^d\left(t\right)\right)+\beta ·}{\left(t\right)+r·\left(x_{i-1}^d\left(t\right)+x_i^d\left(t\right)\right)+\beta ·}}$ ${\displaystyle \genfrac{}{}{0pt}{}{\left(x_{rand}^d\left(t\right)-x_i^d\left(t\right)\right) i=1}{\left(x_{rand}^d\left(t\right)-x_i^d\left(t\right)\right) i=2,\dots N}}$ Foraging in somersault $x_i^d\left(t+1\right)=x_i^d\left(t\right)+S·\left(r2·x_{best}^d-r3·x_i^d\left(t\right)\right),$ $i=1,\dots \dots ..N$ Long term memory Manta Ray foraging optimization $Pi={\displaystyle \frac{f\left(x_i^d\right)}{{\sum }_{j=1}^{ML}f\left(x_j^d\right)}}$ where $x_i^d\left(t\right)$ denotes the position of the $i$-th manta ray in the $d$-th dimension at iteration $t$, $x_{best}^d\left(t\right)$ is the best solution found so far, $N$ is the population size, $r,r_2,r_3\in (0, 1)$ are random numbers, $\alpha$ and $\beta$ are adaptive coefficients controlling exploration and exploitation, $S$ is the somersault factor, $l{b}^d$ and $u{b}^d$ are the lower and upper bounds of the $d$-th variable, $f(\cdot )$ is the fitness function, $ML$ is the memory length, and all position variables are expressed in units consistent with the decision space.
[288]	The study explored the modeling of smart buildings using non-responsive devices and renewable photovoltaic sources, incorporating the KNX protocol for energy management and integrating batteries for energy storage and peak load.	The whale optimization algorithm (WOA) was used.	The study demonstrated that strategic battery charging and discharging management and photovoltaic unit utilization significantly reduced operating costs, as demonstrated through a 30 modified system test system.	$\mathrm{Distance} \mathrm{vector} \overrightarrow{D}=\left|\overrightarrow{C}\cdot \overrightarrow{x^{\ast }}\left(t\right)-\overrightarrow{x}\left(t\right)\right|$ $\mathrm{Position} \mathrm{update} \overrightarrow{x}\left(t+1\right)=\overrightarrow{x^{\ast }-}\overrightarrow{A}·\overrightarrow{D}$ $\mathrm{Coefficient} \mathrm{vector} \overrightarrow{A}=2\overrightarrow{a}·\overrightarrow{r}-\overrightarrow{a}$ $\mathrm{Coefficient} \mathrm{vector} \overrightarrow{C}=2·\overrightarrow{r}$ Total cost function $min\left\{cost={\displaystyle \sum _{t=1}^T}{C}_t\times ∆t\times \left(-E_t^s\right)+{\displaystyle \sum _{y=1}^{Ny}}daycos{t}_y\right\}$ where $\overrightarrow{x}(t)$ and $\overrightarrow{x^{\ast }}(t)$ denote the current and best solution vectors at iteration $t$, respectively, $\overrightarrow{D}$ is the distance vector with units consistent with the decision variables, $\overrightarrow{A}$ and $\overrightarrow{C}$ are dimensionless coefficient vectors controlling exploration and exploitation, $a$ is a linearly decreasing control parameter, and $\overrightarrow{r}$ is a random vector uniformly distributed in $[0,1]$; $C_t$ is the electricity price at time $t$ (currency/kWh), $\mathsf{\Delta }t$ is the time-step duration (h), $E_t^s$ is the scheduled exported or shifted energy at time $t$ (kWh, negative sign indicating cost reduction), ${\mathrm{daycost}}_y$ is the aggregated daily operational cost for day $y$ (currency), $T$ is the total number of time steps, and $N_y$ is the number of days considered.
[289]	This study proposed an improved grasshopper optimization algorithm, termed Outpost Multi-population GOA, which enhances local exploitation and global exploration through outpost and multi-population mechanisms, and experimental results show that it outperforms conventional algorithms in high-dimensional optimization and achieves strong performance in a real-world lithology prediction task.	Outpost Multi-population Grasshopper Optimization Algorithm (OMGOA) was used.	The study showed that the proposed Outpost Multi-population Grasshopper Optimization Algorithm (OMGOA) consistently outperformed standard GOA and other metaheuristic methods in complex and high-dimensional optimization tasks, while also achieving strong and reliable performance in real-world lithology prediction.	Mathematical formula for proposed OMGOA method $\{\begin{array}{c}\left[\lambda \right]=\min \left(\mathrm{function}\left(S_{temp}\right),\mathrm{function}\left(S_i\right)\right)\\ S_i=S_{\lambda }\end{array}$ where $f(\cdot )$ denotes the objective function and $\lambda$ represents the updated elite position.
[290]	Gradient Boosting was used in SHEMS to enhance its intelligence by analyzing complex datasets, detecting patterns, and making data-driven decisions for energy optimization. This enabled it to adapt to dynamic usage patterns, predict future consumption trends, and identify energy savings opportunities.	Gradient Boosting was used.	The Gradient Boosting algorithm outperformed other ML algorithms in predicting energy consumption for smart homes, with a score of 0.95, an RMSE of 6.8, and an MAE of 5.2.	$\mathrm{Initialization}: F_0\left(x\right)=mean\left(y\right)$ Compute Pseudo-Residuals $r_{im}=-\left[{\displaystyle \frac{\partial L\left(y_i,F_{m-1}\left(X_i\right)\right)}{\partial F_{m-1}\left(x_i\right)}}\right]F_{m-1}\left(x\right)=F_{m-1}\left(X_i\right)$ Fit Weak Learner to Pseudo-Residuals: $h_m\left(x\right)=\arg {min}_h{\displaystyle \sum _{i=1}^n}L\left(y_i,F_{m-1}\left(X_i\right)+h\left(X_i\right)\right)$ Update Model: $F_m\left(x\right)=F_{m-1}\left(x\right)+\gamma h_m\left(x\right)$ Output Final Model: $F\left(x\right)={\displaystyle \sum _{m=1}^M}\gamma h_m\left(x\right)$ where $x_i$ and $y_i$ denote the input features and target value of the $i$-th sample, $F_m\left(x\right)$ is the ensemble prediction after $m$ boosting iterations, $L(\cdot )$ is a differentiable loss function, $r_{i,m}$ are the pseudo-residuals at iteration $m$, $h_m\left(x\right)$ is the weak learner fitted to the residuals, $\gamma$ is the learning rate (dimensionless), $n$ is the number of training samples, and $M$ is the total number of boosting stages.
[291]	The power scheduling problem in a smart home (PSPSH) aims to reduce electricity costs, balance power consumption during peak periods, and maximize user satisfaction, but achieving optimal solutions is often limited by specific constraints.	This paper employed the grey wolf optimizer (GWO).	The proposed BMO-PSPSH approach outperforms 17 state-of-the-art algorithms on their datasets and four algorithms on the proposed datasets, achieving superior performance across nearly all power consumption and dynamic pricing scenarios.	$Encircling Prey$ $\mathrm{Distance} \mathrm{vector} \overrightarrow{D}=\left|\overrightarrow{C}\times \overrightarrow{X_p}\left(itr\right)-\overrightarrow{X}\left(itr\right)\right|$ $\mathrm{Position} \mathrm{update} \overrightarrow{X}\left(itr+1\right)=\overrightarrow{X_p}\left(itr\right)-\overrightarrow{A}\times \overrightarrow{D}$ $\mathrm{Solution} \mathrm{set} S=\left[s_{1,}{s}_{2,}\dots ,s_m\right],$ $\mathrm{Neighboring} \mathrm{solution} \mathrm{set} NS=\left[{ns}_{1,}n{s}_{2,}\dots ,{ns}_q\right]$ where $\overrightarrow{X}(\mathrm{itr})$ and $\overrightarrow{X}(\mathrm{itr}+1)$ denote the current and updated positions of a search agent at iteration itr, respectively, ${\overrightarrow{X}}_p(\mathrm{itr})$ represents the best (prey) position found so far, $\overrightarrow{D}$ is the distance vector, and $\overrightarrow{A}$ and $\overrightarrow{C}$ are dimensionless coefficient vectors controlling exploitation and exploration; $S$ denotes the set of candidate solutions, $NS$ is the corresponding neighboring solution set, and all position vectors are expressed in units consistent with the decision variables of the optimization problem.
[292]	This research proposed a novel HEM system that integrates battery energy storage systems (BESSs), PV systems, and electric vehicles (EVs) using an MILP approach to reduce electricity costs.	The research used mixed-integer linear programming (MILP).	The results showed a significant 46.38% reduction in electricity costs for multiple smart homes compared to a traditional scenario without PV, BESS, or EV integration.	$\mathrm{Flexible} \mathrm{load} \mathrm{energy} \mathrm{consumption} E_F^t=\sum _{iϵF}\left[P_{i,F}\times b_{i,F}^t\times ∆t\right], \forall t$ $\mathrm{Non}\text{-}\mathrm{flexible} \mathrm{load} \mathrm{energy} \mathrm{consumption} E_{NF}^t=\sum _{jϵNF}\left[P_{j,NF}\times ∆t\right], \forall t$ Cost minimization objective $min{\displaystyle \sum _{i=1}^N}\left(C_i\right)=min{\displaystyle \sum _{i=1}^N}{\displaystyle \sum _{t=1}^{24}}\left(p^t\times E_{G,i}^t\right), \forall t$ $\mathrm{where} E_F^t$ $\mathrm{and} E_{NF}^t$ $\mathrm{denote} \mathrm{the} \mathrm{total} \mathrm{flexible} \mathrm{and} \mathrm{non}\text{-}\mathrm{flexible} \mathrm{energy} \mathrm{consumption} \mathrm{at} \mathrm{time} t$(kWh), respectively, $P_{i,F}$ and $P_{j,NF}$ $\mathrm{are} \mathrm{the} \mathrm{rated} \mathrm{power} \mathrm{of} \mathrm{flexible} \mathrm{and} \mathrm{non}\text{-}\mathrm{flexible} \mathrm{appliances} (\mathrm{kW}), b_{i,F}^t\in \{0, 1\}$ $\mathrm{is} \mathrm{the} \mathrm{operating} \mathrm{status} \mathrm{of} \mathrm{flexible} \mathrm{appliance} i$ at time $t$, $\mathsf{\Delta }t$ $\mathrm{is} \mathrm{the} \mathrm{time}\text{-}\mathrm{step} \mathrm{duration} \left(\mathrm{h}\right), F$ and $NF$ $\mathrm{represent} \mathrm{the} \mathrm{sets} \mathrm{of} \mathrm{flexible} \mathrm{and} \mathrm{non}\text{-}\mathrm{flexible} \mathrm{loads}, p^t$ $\mathrm{is} \mathrm{the} \mathrm{electricity} \mathrm{price} \mathrm{at} \mathrm{time} t$ $(\mathrm{currency}/\mathrm{kWh}), E_{G,i}^t$ $\mathrm{is} \mathrm{the} \mathrm{grid}\text{-}\mathrm{supplied} \mathrm{energy} \mathrm{for} \mathrm{user} \mathrm{or} \mathrm{appliance} i$ $\mathrm{at} \mathrm{time} t$ (kWh), $C_i$ $\mathrm{is} \mathrm{the} \mathrm{total} \mathrm{energy} \mathrm{cosT} \mathrm{associated} \mathrm{with} \mathrm{index} i$ $\left(\mathrm{currency}\right), N$ $\mathrm{is} \mathrm{the} \mathrm{total} \mathrm{number} \mathrm{of} \mathrm{users}/\mathrm{appliances}, \mathrm{and} t$ spans a 24 h scheduling horizon.
[293]	They aimed to develop a smart home energy management system (HEMS) to efficiently operate residential electrical appliances.	The model used genetic algorithm (GA) optimization, with results demonstrating its effectiveness.	The results highlighted the model’s effectiveness.	$J1=MinCost=Min{\displaystyle \sum _{q=1}^T}PR{c}_q\left(Pq\right)\times Pq\times ∆\mathrm{t}\left(q\right)$ s.t: $Pq=Min {\displaystyle \sum _{i=1}^v}{P}_i\times u_i\left(q\right)$ Water Heater Modeling $C_w={\displaystyle \frac{dThw}{dt}}=Q_{elec}-m{C}_p\left(T_w-T_{inlet}\right)+\left(T_{amb}-T_w\right)$ Air Conditioning System Modeling $\left({\displaystyle \frac{dQ}{dt}}\right)=\left(T_{heater}-T_{room}\right)·M_{dot}·C$ ${\left({\displaystyle \frac{dQ}{dt}}\right)}_{losses}={\displaystyle \frac{\left(T_{heater}-T_{outdoor}\right)}{req}}$ $\left({\displaystyle \frac{d{T}_{room}}{dt}}\right)={\displaystyle \frac{1}{M_{dot}·C}}\left({\displaystyle \frac{d{Q}_{heater}}{dt}}-{\displaystyle \frac{d{Q}_{losses}}{dt}}\right)$ $\mathrm{where} J_1$ $\mathrm{denotes} \mathrm{the} \mathrm{total} \mathrm{electricity} \mathrm{cost} \left(\mathrm{currency}\right), PR{c}_q(\cdot )$ $\mathrm{is} \mathrm{the} \mathrm{electricity} \mathrm{price} \mathrm{function} \mathrm{at} \mathrm{time} \mathrm{slot} q$ $(\mathrm{currency}/\mathrm{kWh}), P_q$ $\mathrm{is} \mathrm{the} \mathrm{aggregated} \mathrm{electrical} \mathrm{power} \mathrm{demand} \mathrm{at} \mathrm{time} q$(kW), $\mathsf{\Delta }{t}_q$ $\mathrm{is} \mathrm{the} \mathrm{duration} \mathrm{of} \mathrm{time} \mathrm{slot} q$(h), $P_i$ $\mathrm{is} \mathrm{the} \mathrm{rated} \mathrm{power} \mathrm{of} \mathrm{appliance} i$(kW), and $u_i\left(q\right)\in \{0, 1\}$ $\mathrm{is} \mathrm{its} \mathrm{operating} \mathrm{status}; C_w$ $\mathrm{is} \mathrm{the} \mathrm{thermal} \mathrm{capacitance} \mathrm{of} \mathrm{the} \mathrm{water} \mathrm{heater} (\mathrm{kJ}/°\mathrm{C}), T_{hw}$, $T_w$, $T_{\mathrm{inlet}}$, and $T_{\mathrm{amb}}$ $\mathrm{denote} \mathrm{hot} \mathrm{water}, \mathrm{tank}, \mathrm{inlet}, \mathrm{and} \mathrm{ambient} \mathrm{temperatures} (°\mathrm{C}), Q_{\mathrm{elec}}$ $\mathrm{is} \mathrm{the} \mathrm{electrical} \mathrm{heating} \mathrm{power} \left(\mathrm{kW}\right), m$ $\mathrm{is} \mathrm{the} \mathrm{water} \mathrm{mass} \mathrm{flow} \mathrm{rate} (\mathrm{kg}/\mathrm{s}), \mathrm{and} C_p$ $\mathrm{is} \mathrm{the} \mathrm{specific} \mathrm{heat} \mathrm{capacity} \mathrm{of} \mathrm{water} (\mathrm{kJ}/\mathrm{kg}·°\mathrm{C}); T_{\mathrm{heater}}$, $T_{\mathrm{room}}$, and $T_{\mathrm{outdoor}}$ $\mathrm{are} \mathrm{heater}, \mathrm{indoor}, \mathrm{and} \mathrm{outdoor} \mathrm{temperatures} (°\mathrm{C}), \dot{M}$ $\mathrm{is} \mathrm{the} \mathrm{air} \mathrm{mass} \mathrm{flow} \mathrm{rate} (\mathrm{kg}/\mathrm{s}), C$ $\mathrm{is} \mathrm{the} \mathrm{specific} \mathrm{heat} \mathrm{capacity} \mathrm{of} \mathrm{air} (\mathrm{kJ}/\mathrm{kg}·°\mathrm{C}), \mathrm{and} R_{eq}$ is the equivalent thermal resistance of the building envelope (°C/kW).
[294]	This study aimed to solve the PSPSH by minimizing electricity bills, improving user comfort, and maintaining power system performance using the GWO-MCA method. Its impact on five other optimization algorithms was also evaluated.	This paper combined the Min-Conflict Local Search Algorithm (MCA) with the grey wolf optimizer (GWO).	GWO-MCA outperformed all compared MCA-based methods and three state-of-the-art hybrid methods in solving PSPSH. It also outperforms 20 other state-of-the-art methods across most datasets.	GWO-MCA $A_l=2\times a_l\times r_l-a_{l,}$ $a_l=2-\left(2\times {\displaystyle \frac{itr}{I}}\right)$ $S=\left[s_{1,}{s}_{2,}\dots ,s_m\right],$ $T=\left[t^1,t^2,\dots ,t^n\right]$ $\mathrm{where} A_l$ $\mathrm{denotes} \mathrm{the} \mathrm{position} \mathrm{update} \mathrm{for} \mathrm{agent} l$, $a_l$ $\mathrm{is} \mathrm{the} \mathrm{scaling} \mathrm{factor} \mathrm{that} \mathrm{decreases} \mathrm{over} \mathrm{iterations}, r_l$ $\mathrm{is} \mathrm{the} \mathrm{random} \mathrm{component} \mathrm{generated} \mathrm{within} \mathrm{the} \mathrm{range} [0, 1], \mathrm{and} a_{l-1}$ $\mathrm{represents} \mathrm{the} \mathrm{scaling} \mathrm{factor} \mathrm{from} \mathrm{the} \mathrm{previous} \mathrm{iteration}. itr$ $\mathrm{is} \mathrm{the} \mathrm{current} \mathrm{iteration} \mathrm{number}, I$ $\mathrm{is} \mathrm{the} \mathrm{total} \mathrm{number} \mathrm{of} \mathrm{iterations}, \mathrm{and} S=[s_1,s_2,\dots ,s_m]$ $\mathrm{represents} \mathrm{the} \mathrm{set} \mathrm{of} \mathrm{solutions} \mathrm{or} \mathrm{positions} \mathrm{in} \mathrm{the} \mathrm{search} \mathrm{space}. T=[t^1,t^2,\dots ,t^n]$ denotes a set of thresholds or criteria used during the optimization process.
[295]	This study aimed to develop cities that face increasing uncertainty, necessitating smart devices and apps for security, requiring strict measures to protect personal information and prevent illegal access, and requiring reliable computing for the IoT.	The whale optimization algorithm with deep convolutional neural networks was used.	This research showcased the effectiveness of the proposed approach in defending smart home systems from safety risks, advancing IoT security in a growing connected world.	$\mathrm{Encircling}: \overrightarrow{I}\left(t+1\right)=\overrightarrow{I}\times \left(t\right)-\overrightarrow{A}\times \overrightarrow{D}$ $\overrightarrow{D}=\left(\overrightarrow{C}\times \overrightarrow{I}\times \left(t\right)-\overrightarrow{I}\left(t\right)\right)$ $\overrightarrow{A}=2\times \overrightarrow{a}\times \overrightarrow{r}-\overrightarrow{a}$ $\overrightarrow{C}=2\times \overrightarrow{r}$ $\mathrm{Hunting}: \overrightarrow{I}\left(t+1\right)=\overrightarrow{D}\times e^{hl}\times cos\left(2\pi l\right)+\overrightarrow{I}$ $\overrightarrow{I}\left(t+1\right)=\left\{{\displaystyle \genfrac{}{}{0pt}{}{ \overrightarrow{I}\times \left(t\right)-\overrightarrow{A}\times \overrightarrow{D} if p<0·5 }{\overrightarrow{D}\times e^{hl}\times \cos \left(2\pi l\right)+\overrightarrow{I}\times if p<0·5}}\right\}$ Searching: $\overrightarrow{D}=\left(\overrightarrow{C}\times {\overrightarrow{I}}_{rand}-\overrightarrow{I}\right)$ $\overrightarrow{I}\left(t+1\right)=\overrightarrow{I_{rand}}-\overrightarrow{A}\times \overrightarrow{D}$ $\mathrm{where} \overrightarrow{I}(t)$ $\mathrm{and} \overrightarrow{I}(t+1)$ $\mathrm{denote} \mathrm{the} \mathrm{current} \mathrm{and} \mathrm{updated} \mathrm{solution} \mathrm{vectors} \mathrm{at} \mathrm{iteration} t$, $\overrightarrow{I^{\ast }}(t)$ $\mathrm{represents} \mathrm{the} \mathrm{best} (\mathrm{leader}) \mathrm{solution}, \mathrm{and} {\overrightarrow{I}}_{rand}$ $\mathrm{is} \mathrm{a} \mathrm{randomly} \mathrm{selected} \mathrm{solution}; \overrightarrow{A}$ $\mathrm{and} \overrightarrow{C}$ $\mathrm{are} \mathrm{coefficient} \mathrm{vectors} \mathrm{controlling} \mathrm{exploitation} \mathrm{and} \mathrm{exploration}, a_l$ $\mathrm{is} \mathrm{a} \mathrm{linearly} \mathrm{decreasing} \mathrm{control} \mathrm{parameter}, \overrightarrow{r}$ $\mathrm{and} r_l$ $\mathrm{are} \mathrm{random} \mathrm{vectors} \mathrm{uniformly} \mathrm{distributed} \mathrm{in} [0,1]$, $p$ $\mathrm{is} \mathrm{a} \mathrm{probability} \mathrm{threshold}, h$ $\mathrm{and} l$ $\mathrm{are} \mathrm{spiral}\text{-}\mathrm{motion} \mathrm{parameters}, I$ $\mathrm{is} \mathrm{the} \mathrm{maximum} \mathrm{number} \mathrm{of} \mathrm{iterations}, \mathrm{and} \mathrm{all} \mathrm{vectors} \mathrm{are} \mathrm{expressed} \mathrm{in} \mathrm{units} \mathrm{consistent} \mathrm{with} \mathrm{the} \mathrm{decision} \mathrm{variables}; S$ $\mathrm{denotes} \mathrm{the} \mathrm{solution} \mathrm{set} \mathrm{and} T$ the associated evaluation criteria used in the multi-criteria decision framework.
[296]	This study aimed to optimize energy management in smart buildings with electric vehicles by considering risk, economic factors, and practical constraints like production limits and flexible loads.	The whale optimization algorithm was used.	The results showed that a positive consumer attitude reduced net costs, with the battery used during favorable price periods. Price risk indicators also impact users’ electricity purchasing strategies differently.	Besieging the prey: $\overrightarrow{D}=\left|\overrightarrow{C}·\overrightarrow{X^{\ast }}\left(t\right)-\overrightarrow{X_{\left(t\right)}}\right|$ $\overrightarrow{X}\left(t+1\right)=\overrightarrow{X^{\ast }}\left(t\right)-\overrightarrow{A}·\overrightarrow{D}$ $\overrightarrow{A}=2\overrightarrow{a}·\overrightarrow{r}-\overrightarrow{a}$ $\overrightarrow{C}=2·\overrightarrow{r}$ Bubble attack calculation (operation phase): $\overrightarrow{X}\left(t+1\right)=D^{\to }.e^{bl}.\cos \left(2\pi l\right)+\overrightarrow{X^{\ast }}\left(t\right)$ $\overrightarrow{X}\left(t+1\right)=\left\{\begin{array}{c}\overrightarrow{X^{\ast }}\left(t\right)-\overrightarrow{A}.\overrightarrow{D} if p<0.5\\ D^{\to }.e^{bl}.\cos \left(2\pi l\right)+\overrightarrow{X^{\ast }}\left(t\right) if p>0.5\end{array}\right\}$ Searching for the prey (exploration phase): $\overrightarrow{D}=\left|\overrightarrow{C}.\overrightarrow{X_{rand}}-\overrightarrow{X}\right|$ $\overrightarrow{X}\left(t+1\right)=\overrightarrow{X_{rand}}-\overrightarrow{A}·\overrightarrow{D}$ $\left|P_{grid}\left(t\right)\right|\le P_{grid,max}\left(t\right)$ $t_{load,start}\left(t\right)\le t_{load}\le t_{load,end}-N,tϵN^{\ast }$ $0\le P_{air}\left(t\right)\le P_{air}, max$ $T_{air,}min\le T_{in}\le T_{air,}max$ $\mathrm{where} \overrightarrow{X}(t)$ $\mathrm{and} \overrightarrow{X}(t+1)$ $\mathrm{denote} \mathrm{the} \mathrm{current} \mathrm{and} \mathrm{updated} \mathrm{whale} \mathrm{positions} \mathrm{at} \mathrm{iteration} t$ $, \overrightarrow{X^{\ast }}(t)$ $\mathrm{is} \mathrm{the} \mathrm{best} \mathrm{solution} \mathrm{found} \mathrm{so} \mathrm{far}, {\overrightarrow{X}}_{rand}$ $\mathrm{is} \mathrm{a} \mathrm{randomly} \mathrm{selected} \mathrm{solution}, \mathrm{and} \overrightarrow{D}$ $\mathrm{is} \mathrm{the} \mathrm{distance} \mathrm{vector}; \overrightarrow{A}$ $\mathrm{and} \overrightarrow{C}$ $\mathrm{are} \mathrm{dimensionless} \mathrm{coefficient} \mathrm{vectors}, \overrightarrow{a}$ $\mathrm{is} \mathrm{a} \mathrm{linearly} \mathrm{decreasing} \mathrm{control} \mathrm{parameter}, \overrightarrow{r}$ $\mathrm{is} \mathrm{a} \mathrm{random} \mathrm{vector} \mathrm{uniformly} \mathrm{distributed} \mathrm{in} [0,1]$, $p$ $\mathrm{is} \mathrm{a} \mathrm{probability} \mathrm{threshold}, \mathrm{and} b$ $\mathrm{and} l$ $\mathrm{are} \mathrm{spiral}\text{-}\mathrm{shape} \mathrm{parameters}; P_{grid}\left(t\right)$ $\mathrm{is} \mathrm{grid} \mathrm{power} \mathrm{exchange} (\mathrm{kW}), P_{grid,\mathrm{m}\mathrm{a}\mathrm{x}}(t)$ $\mathrm{is} \mathrm{its} \mathrm{maximum} \mathrm{allowable} \mathrm{value} \left(\mathrm{kW}\right), P_{air}\left(t\right)$ $\mathrm{is} \mathrm{air}\text{-}\mathrm{conditioning} \mathrm{power} \left(\mathrm{kW}\right), T_{in}$ is indoor temperature (°C), and all position vectors are expressed in units consistent with the decision variables.

shows that the results from the various studies on smart home energy management optimization reveal several key findings about the effectiveness of different methods. Across the board, WOA, the grey wolf optimizer (GWO), and the cuckoo optimization algorithm (COA) demonstrate strong performance in reducing energy costs and improving system efficiency. These algorithms, when applied to real-time energy management scenarios, show substantial reductions in grid reliance and electricity expenses. For instance, WOA achieves a 46.6% reduction in grid dependence and a 57.7% decrease in energy costs, highlighting its ability to dynamically adapt to changing energy demands. Similarly, GWO is found to excel in integrating renewable energy sources (RESs) into energy management systems, further reducing energy consumption and costs. In comparison, the cuckoo optimization algorithm, particularly in scenarios that involve distributed renewable energy systems, delivers impressive savings ranging from 57% to 80%, emphasizing the importance of optimizing energy load scheduling in conjunction with RESs. Other algorithms, such as the improved sine cosine algorithm (ISCA), outperform traditional methods like grasshopper optimization algorithm (GOA), achieving greater energy cost reductions and improving the peak-to-average ratio (PAR), which is critical for balancing grid usage during peak times. These findings show that the balance between exploration and exploitation within these optimization algorithms is vital for effective energy management. Additionally, user comfort remains a central consideration, with many algorithms, including Adaptive Coati Optimization, enhancing user satisfaction while optimizing energy usage, showing that it is not just about cost reduction but also about maintaining a comfortable environment for residents. The integration of electric vehicles (EVs), PV systems, and battery storage in some studies shows further potential for optimizing energy usage and reducing costs, while ensuring that RESs are maximized. The use of hybrid and multi-objective optimization approaches in some cases, like Reinforced Learning Quantum Inspired Grey Wolf Optimization (RLQIGWO), introduces a layer of adaptability that is essential for systems operating in real time with fluctuating energy demands, further supporting the trend toward more flexible and sustainable energy management systems. Overall, the studies reveal that combining these optimization methods with renewable energy integration and real-time energy management strategies provides a comprehensive solution to the challenges of modern energy consumption in smart homes.

6. Conclusions

This review presents the integration of artificial intelligence (AI) models in optimizing energy management within smart homes, with a particular focus on clean and low-carbon technologies. The study highlights the significant potential of AI-driven solutions in enhancing energy efficiency, reducing carbon footprints, and improving overall energy management in smart homes. Key findings from this review demonstrate the considerable advantages of these advanced technologies in addressing the challenges of modern energy systems, which are outlined as follows:

• Machine learning (ML), deep learning, and heuristic algorithms have proven essential for enabling predictive energy management, demand-side optimization, and adaptive scheduling in smart homes. Specifically, ML models like Gradient Boosting were shown to significantly improve energy consumption predictions, achieving high accuracy scores (>0.95) in several studies. These models illustrate the potential of ML in enhancing energy efficiency by predicting demand and optimizing usage in real-time.
• The integration of renewable energy sources with AI-optimized energy storage systems plays a crucial role in improving sustainability by reducing dependence on conventional power grids. In approximately 40% of the reviewed studies, energy storage systems paired with renewable energy sources demonstrated substantial savings and increased operational efficiency. AI-powered HEMSs enable real-time monitoring, energy forecasting, and adaptive control, leading to both reduced energy costs and enhanced user comfort.
• Optimization algorithms, such as genetic algorithms (GAs), particle swarm optimization (PSO), and mixed-integer linear programming (MILP), have been widely applied across various energy management tasks, from load scheduling to multi-objective optimization. Among the optimization techniques reviewed, WOA stood out in 40% of the studies for its remarkable ability to reduce grid reliance by 46.6% and energy costs by 57.7% in real-time applications. This highlights WOA’s significant impact on dynamic, real-time energy management. Similarly, the cuckoo optimization algorithm (COA), used predominantly in renewable energy systems, yielded energy cost savings between 57% and 80%, appearing in roughly 30% of the studies. Furthermore, the Adaptive Coati Optimization Algorithm demonstrated its ability to balance cost reduction with improved comfort, showing a 20% increase in user satisfaction while reducing electricity costs.
• The incorporation of blockchain technology for peer-to-peer (P2P) energy trading, combined with IoT-assisted smart grids and decentralized energy systems, further improves the efficiency and scalability of energy management solutions in smart homes. These technologies empower consumers and enable the more efficient distribution of energy across networks.

Despite these advancements, several challenges persist, particularly in real-time energy management, data integration, and computational complexity. The high computational costs of certain optimization techniques, such as MILP, limit their scalability and practical application in dynamic energy systems. Furthermore, the integration of AI with existing infrastructure remains a significant hurdle due to the lack of standardization and the complexity of achieving effective system compatibility. To address the challenges in AI-driven energy optimization, future research should focus on developing lightweight AI models to alleviate computational burdens, improve data availability and quality, and enhance system integration across diverse platforms. Additionally, the scalability of AI solutions, particularly in large-scale residential environments, needs to be prioritized, as current studies often fail to account for integration challenges with existing infrastructures, such as traditional grids and older systems.