Artificial Intelligence in Solar-Assisted Greenhouse Systems: A Technical, Systematic and Bibliometric Review of Energy Integration and Efficiency Advances

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

1. The search query in the article excludes ("greenhouse gas*" OR "greenhouse effect"). Could this exclusion lead to the omission of some engineering papers that discuss the mitigation of greenhouse gases through AI-driven solar systems? It is recommended that the authors explain this exclusion criterion or conduct a cross-verification to ensure that relevant technical literature discussing emission reduction effects has not been overlooked.
2. Several bibliometric network maps (e.g., Figures 6, 7, 9, and 11) appear to have low resolution, and Figure 11 suffers from overlapping small labels. It is recommended to provide high-resolution images or vector graphics to ensure that the content of these charts is clearly legible.
3. Regarding the organization of the sections, Section 3.13 is titled "Artificial Intelligence applied to..." , while Section 3.15 is titled "Machine Learning and solar energy...". Given that Machine Learning (ML) is a core subset of AI, and there is significant logical and content overlap between these two sections (for instance, Section 3.13 also discusses Neural Networks ), it is recommended to either merge these sections or more clearly delineate their respective dimensions of discussion to avoid redundancy.
4. If the data retrieval did not cover the full year of 2025, it is suggested to indicate in the caption of Figure 2 or in the Methodology section that the "2025 data is partial" or "Year-to-date."
5. There is a distinct layout error in the manuscript: Figure 11 is duplicated on both page 16 and page 17. Please remove the redundant figure in the final version.

Author Response

Bogota December 04 2025.

Reviewer 1 Technologies.

Dear. Reviewer.

We sincerely thank Reviewer 1 for the thorough and insightful comments provided. Your observations significantly strengthened the clarity, structure, and analytical depth of the manuscript. All suggested modifications have been carefully addressed, and the corresponding changes are highlighted in green in the revised version. A detailed explanation of each revision is also included in this response letter. We appreciate the time and expertise invested in reviewing our work, which has undoubtedly improved the overall quality of the paper.

 

Comments.

The search query in the article excludes ("greenhouse gas*" OR "greenhouse effect"). Could this exclusion lead to the omission of some engineering papers that discuss the mitigation of greenhouse gases through AI-driven solar systems? It is recommended that the authors explain this exclusion criterion or conduct a cross-verification to ensure that relevant technical literature discussing emission reduction effects has not been overlooked.

Reply. Thank you very much for this thoughtful observation. We agree that the exclusion of the terms "greenhouse gas*" and "greenhouse effect" requires explicit justification, given the potential link with emission-mitigation studies. In our original search strategy, this exclusion criterion was introduced to avoid many records in which the word “greenhouse” refers to atmospheric greenhouse gases and global climate models rather than to protected-crop structures (greenhouses, solar dryers, agrivoltaic greenhouses, etc.). Preliminary tests showed that, without this filter, the query retrieved a substantial volume of articles focused on macro-scale climate policy, atmospheric chemistry, or power systems with no explicit connection to controlled agricultural environments. Following your suggestion, we have now re-run the search in Scopus without this exclusion and manually screened the additional records. The extra documents were predominantly related to economy-wide emission scenarios, carbon pricing, or large-scale energy systems. None of them fulfilled our predefined inclusion criteria, which required a clear focus on: (i) greenhouses, solar dryers, or agrivoltaic structures, and (ii) the application of AI/ML/IoT to solar-assisted energy and microclimate management in those systems. Therefore, the final corpus of 79 articles remains unchanged.

The following text was entered in section 2.2:

The results were exported in CSV and RIS formats for management and processing in Excel and Mendeley. The exclusion clause NOT ("greenhouse gas*" OR "greenhouse effect") was introduced to filter out studies in which the term “greenhouse” refers to atmospheric greenhouse gases or the planetary greenhouse effect, rather than to protected-crop structures or solar dryers. This prevented the retrieval of many macro-scale climate or energy-policy studies outside the scope of controlled agricultural environments. To ensure that this filter did not bias the dataset against relevant engineering works on emission mitigation, we re-ran the query without this exclusion and manually screened the additional records. The newly retrieved documents mainly addressed economy-wide or power-system-level emission scenarios and did not meet the predefined inclusion criteria; consequently, the final corpus of 79 articles remained un-changed.

 

Several bibliometric network maps (e.g., Figures 6, 7, 9, and 11) appear to have low resolution, and Figure 11 suffers from overlapping small labels. It is recommended to provide high-resolution images or vector graphics to ensure that the content of these charts is clearly legible.

Reply. We appreciate this helpful observation and fully agree that the bibliometric network maps must be clearly legible. All network figures (Figures 6, 7, 9, and 11) were originally generated as vector graphics (SVG). Following your suggestion, we have now re-exported all of them as high-resolution vector files (SV) and 600 dpi raster versions, and replaced the low-resolution images in the manuscript.In addition, Figure 11 has been redesigned to avoid overlapping labels by (i) increasing the plotting area, (ii) displaying labels only for the most relevant terms (based on minimum occurrence thresholds). The revised manuscript now includes these improved figures, which are clearly legible both on screen and in print.

 

Regarding the organization of the sections, Section 3.13 is titled "Artificial Intelligence applied to..." , while Section 3.15 is titled "Machine Learning and solar energy...". Given that Machine Learning (ML) is a core subset of AI, and there is significant logical and content overlap between these two sections (for instance, Section 3.13 also discusses Neural Networks ), it is recommended to either merge these sections or more clearly delineate their respective dimensions of discussion to avoid redundancy.

Reply. We sincerely thank the reviewer for this insightful observation. We fully agree that Machine Learning (ML) is a core subset of Artificial Intelligence (AI), and that the original structure of Sections 3.13 and 3.15 generated an undesirable overlap in concepts, methodologies, and applications. To address this, we have merged both sections into a single, integrated and logically coherent section titled “Artificial Intelligence and Machine Learning for Solar and Climate Management in Controlled Agriculture: Optimization, Prediction, and Predictive Sustainability.”

The new section remained as:

3.13. Artificial Intelligence and Machine Learning for Solar and Climate Management in Con-trolled Agriculture: Optimization, Prediction, and Predictive Sustainability

Artificial Intelligence (AI) and its core subset, Machine Learning (ML), are re-shaping the management of solar energy and microclimate in smart greenhouses by enabling autonomous optimization of environmental conditions, resource use, and en-ergy efficiency. Through the integration of microclimatic, agronomic, and energy data, AI improves predictive capacity, regulates processes such as irrigation, ventilation, lighting, and solar drying, and strengthens the adaptive capacity of protected agricul-ture in alignment with the Sustainable Development Goals (SDGs). When combined with photovoltaic (PV), thermal, and Internet of Things (IoT) systems, AI and ML support the transition toward low-carbon, resilient, and high-efficiency greenhouse production models.

 Shahbaz et al. [42], applied AI-based multicriteria decision models including quantum theory, M-SWARA, MOORA, and pictorial fuzzy sets to prioritize renewable energy investments, identifying hybrid systems and PV greenhouses as the most sus-tainable alternatives (weights: 0.267 for cost; 0.266 for recycled materials). Similarly, Banluesapy et al. [23] developed a solar-powered smart irrigation system controlled through a Random Forest algorithm with 99.3% accuracy, achieving reductions of 50% in water and energy consumption, a 130% increase in water-use efficiency, and a 50% re-duction in CO₂ emissions. Yu et al. [56], validated a photoelectric drip irrigation system integrating AI and solar energy, improving both water and energy efficiency under greenhouse conditions.

In advanced monitoring and control, Touhami et al. [76], proposed a hybrid archi-tecture based on wireless sensor networks (WSN), IoT, and renewable energies for hy-droponic systems capable of automatically regulating pH, temperature, and water level. Maraveas [77], emphasized that despite notable productivity gains enabled by AI and IoT, barriers persist regarding technological cost, sensor accuracy, and adoption in the Global South. In parallel, commercial implementations from Blue River Technology and John Deere illustrate the growing feasibility of autonomous irrigation, pest control, and energy management systems, although long-term reliability and operational training remain critical challenges.

Relevant advances in solar-energy modeling have also been documented. Harrou et al. [29] synthesized eleven studies and highlighted that hybrid recurrent neural net-works (RNN) and multilayer perceptrons (MLP) achieve correlations above 0.95 (RMSE = 0.19) for photovoltaic forecasting. Algorithms such as XGBoost, Random Forest, and Reinforcement Learning demonstrated robust performance in predicting irradiance and correcting partial shading, while the Temporal Fusion Transformer (TFT) outperformed ARIMA and LSTM across multiscale forecasting tasks. Bezari et al. [25], developed an ANN (6–15–1) yielding correlations above 96% for irradiance estimation in greenhouses, offering valuable inputs for calibrating CFD simulations and adaptive solar-control systems.

In solar drying, Hoque et al. [78], demonstrated that the integration of remote sensing and AI reduces drying time by 25–30%, preserves up to 90% of nutritional content, and decreases energy consumption and emissions by 15–20%. IoT-blockchain architectures further improved traceability and reduced operational costs by 10–15%. ML-assisted optimization of materials also shows promise: Ali et al. [43] reported that a graphene-based K-structured solar absorber optimized through ML reached a spectral absorption above 97% (R² = 0.996; RMSE = 3.2×10⁻⁵). For predictive maintenance, Iqbal et al. [79] combined electrical impedance spectroscopy with ML, achieving F1-scores of 0.975 for dust and microcrack detection and 0.856 for thermal variation identification.

In the same way Choubey et al. [80] emphasized that combining phase change materials (PCM) and nanoparticles increases thermal conductivity and reduces drying time, while algorithms such as RBF, MLP, ANN, and SVM provide accurate predictions of moisture loss and quality metrics (RMSE = 0.99 for temperature; 33.67 for dry mass), tripling the net benefit compared with traditional dryers. Saeidirad et al. [81] and Karaağaç et al. [28] confirmed that integrating AI with multicriteria decision techniques (TOPSIS) optimizes temperature, airflow, and humidity for high-value crops such as saffron and mushrooms. In a CPV/T solar system with nanostructured PCM (Al₂O₃–paraffin), the ANN–SVM approach achieved R² > 0.9 and thermal/exergetic efficiencies of 20% and 8%, respectively.

ML and deep learning have also enhanced sensor communication and energy forecasting. Alturif et al. [45] integrated deep convolutional neural networks (DCNN) with Lagrangian optimization to reduce communication energy in climatic sensor networks. Venkatesan & Cho (2024) showed that a Bi-LSTM model surpassed CNN, LSTM, and GRU architectures (R² = 0.9243; RMSE = 0.0048) in forecasting solar energy consumption up to two weeks ahead. These advancements support dynamic load scheduling and optimized PV production under varying environmental conditions. In the control domain, Hu & You. [52] designed a robust model predictive control framework with AI (RMPC-AI) for photovoltaic-controlled environment agriculture (CEA-PV), improving energy efficiency by 15.4% and reducing water and light pollution by 8.7% and 3.6%, respectively. Venkateswaran & Cho. [49] proposed a hybrid SSA-CNN-LSTM model achieving MAE values of 0.1202 (hourly) and 0.1774 (daily), enhancing energy planning and load distribution in solar greenhouses.

Finally, the integration of energy, water, and emissions is exemplified by the EWMS system of Fiesta & Tria. [54], which uses an Extreme Learning Machine (ELM) to forecast photovoltaic production and crop water requirements, increasing solar utilization by 94.3% and reducing grid consumption by 3.49 kWh. Khani et al. [55] optimized a solar polygeneration system with CO₂ capture through Genetic Programming and RNA, reducing costs by 11.4% and environmental impact by 34.31%. Alsafasfeh et al. [26], implemented a spatial search neural algorithm that increased PV power output and outperformed conventional optimization methods. Together, these studies delineate the transition toward intelligent, predictive, and highly integrated protected agriculture systems, in which AI and ML constitute fundamental pillars of energy optimization, climate resilience, and sustainability.

If the data retrieval did not cover the full year of 2025, it is suggested to indicate in the caption of Figure 2 or in the Methodology section that the "2025 data is partial" or "Year-to-date."

Reply. Thank you for this valuable clarification. Indeed, the bibliometric data for 2025 correspond to the records available in Scopus at the time of data retrieval and therefore represent year-to-date (partial) data, not the full publication output for 2025.

 

There is a distinct layout error in the manuscript: Figure 11 is duplicated on both page 16 and page 17. Please remove the redundant figure in the final version.

Reply. We appreciate the reviewer for pointing out this layout issue. The duplication of Figure 11 on pages 16 and 17 was an unintended formatting error. The redundant copy has now been removed, and only the correct single instance of Figure 11 is retained in the revised manuscript.

 

Regards

The author

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript has conducted a review on artificial intelligence in solar-assisted greenhouse systems. Minor revision should be performed.

Specific comments:

(1) The fourth section could be removed. Because the third section had already contained discussion.

(2) Figure 1 could be improved.

(3) Some mainly quantitative results should be present in the Conclusions section.  

Author Response

Bogota December 03 2025.

Reviewer 2 Technologies.

Dear.

We sincerely appreciate your constructive and well-founded comments. Following your recommendations, we have incorporated your suggestions into the specific sections indicated and substantially improved all graphical materials. The revised figures were regenerated using professional scientific-plotting software to enhance readability, accuracy, and visual consistency. All modifications are clearly marked in the updated manuscript, and we thank you for the valuable insights that contributed to strengthening the quality and scientific rigor of our work.

The fourth section could be removed. Because the third section had already contained discussion.

Reply. We thank the reviewer for this helpful suggestion. Following your recommendation, the original Section 4 has been removed and its relevant content integrated into the consolidated discussion presented in Section 3. This improves the coherence of the manuscript and avoids redundancy between sections.

Figure 1 could be improved.

Reply. Thank you for this constructive suggestion. Figure 1 has now been fully redesigned to enhance clarity and visual alignment. The updated version fits on a single page, features properly aligned boxes, and presents the workflow in a cleaner and more coherent layout. We believe the revised figure significantly improves readability and overall presentation.

Some mainly quantitative results should be present in the Conclusions section.  

The bibliometric analysis confirms a rapid consolidation of the field, with scientific output increasing from 3 publications in 2020 to 27 in 2025, and a total corpus of 79 studies included in this review. Engineering (55 documents), Computer Science (29), and Energy (19) emerged as the dominant subject areas, while China (11 publications) and the United Kingdom (5) led global contributions. Keyword co-occurrence revealed “solar energy” and “solar power generation” as the most frequent terms (10 occurrences each), underscoring the centrality of energy optimization and AI-driven modeling. These indicators demonstrate that AI-assisted solar systems in protected agriculture have reached an important stage of scientific maturity and thematic coherence.

Regards

The authors

 

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript entitled “Artificial Intelligence in Solar-Assisted Greenhouse Systems: A Technical, Systematic and Bibliometric Review of Energy Integration and Efficiency Advances” has been reviewed. The work is interesting and relevant; however, a major revision is needed prior to publication.

1. Abstract needs revision and should be shortened. Begin with the problem statement followed by objectives, methodology in short. Finally, mention only the cardinal findings briefly.
2. Include a nomenclature/ abbreviation table.
3. The Introduction is short and needs revision. It should ideally begin with a broader context leading readers to more specific aspects. Briefly describe all the aspects such as thermal storage, phase change materials, agrovoltaics, …
4. Refer and discuss advances in thermal storage employing different techniques (https://link.springer.com/article/10.1007/s11356-023-31718-8, https://www.sciencedirect.com/science/article/abs/pii/S2451904924003354)
5. Redevelop all the graphs using a professional graphing software.
6. Section 3.4; Line 311: What is the reason behind more evenly distributed production in Europe?
7. Section 3.11: More studies on Building-Integrated Photovoltaics should be discussed as it is a high potential pathway.
8. A table reflecting significant studies on the synergy among IoT, machine learning, and renewable energies for climate-resilient agricultural ecosystems should be added.
9. There is no need for Section 4. Remove it.
10. Improve the Conclusions section. The Conclusions section should indicate research gaps and research directions identified as the results of research presented. Explain the significance of your work more clearly and explicitly.
11. There are a few grammatical and structural errors. The whole manuscript should be revised to improve the language.

Comments on the Quality of English Language

English needs improvement

Author Response

Bogota 04 December

Dear.

We sincerely thank Reviewer 3 for the thorough and constructive comments provided. All suggestions including improvements to structure, clarity, language quality, and the inclusion of additional analytical elements have been fully incorporated into the revised manuscript. The corresponding modifications are clearly highlighted in purple in the updated version to facilitate verification. We appreciate the reviewer’s careful assessment, which has significantly enhanced the rigor, coherence, and overall scientific quality of this work.

Reviewer 3.

Abstract needs revision and should be shortened. Begin with the problem statement followed by objectives, methodology in short. Finally, mention only the cardinal findings briefly.

Reply. We sincerely thank the reviewer for this valuable observation. In accordance with the suggestion, the Abstract has been fully revised and shortened to improve clarity, structure, and scientific precision. The new version now begins with a concise problem statement, clearly presents the aim of the study, and summarizes the methodological approach (bibliometric + systematic review under PRISMA 2020) in a compact way. Only the essential quantitative results and the principal findings of the review are retained, and all acronyms have been defined upon first use to ensure readability.

The new version is:

Protected agriculture increasingly requires solutions that reduce energy consumption and environmental impacts while maintaining stable microclimatic conditions. The integration of Artificial Intelligence (AI), Machine Learning (ML), and Deep Learning (DL) with solar technologies has emerged as a pathway toward autonomous and energy-efficient greenhouses and solar dryers. This study analyzes the scientific and technological evolution of this convergence using a mixed review approach—bibliometric and systematic—following PRISMA 2020 guidelines. From Scopus records (2012–2025), 115 documents were screened and 79 met the inclusion criteria. Bibliometric results reveal accelerated growth since 2019, led by Engineering, Computer Science, and Energy, with China, India, Saudi Arabia, and the United Kingdom as dominant contributors. Thematic analysis identifies four major research fronts: (i) thermal modeling and energy efficiency, (ii) predictive control and microclimate automation, (iii) integration of photovoltaic–thermal (PV/T) systems and phase change materials (PCM), and (iv) sustainability and agrivoltaics. Systematic evidence shows that AI-, ML-, and DL-based models improve solar forecasting, microclimate regulation, and energy optimization; model predictive control (MPC), deep reinforcement learning (DRL), and energy management systems (EMS) enhance operational efficiency; and PV/T–PCM hybrids strengthen heat recovery and storage. Remaining gaps include long-term validation, metric standardization, and cross-context comparability. Overall, the field is advancing toward near-zero-energy greenhouses powered by Internet of Things (IoT), AI, and solar energy, enabling resilient, efficient, and decarbonized agro-energy systems.

Include a nomenclature/ abbreviation table.

Reply. We thank the reviewer for this valuable suggestion. In response, we have incorporated a comprehensive Table of Abbreviations to enhance clarity and improve the readability of the manuscript, given the extensive use of technical terminology related to Artificial Intelligence, photovoltaic systems, thermal storage, and predictive control algorithms. The new table consolidates all acronyms used throughout the paper (Sections 1–4), providing standardized definitions that facilitate interpretation for both specialized and non-specialized readers.

Abbreviations

The following abbreviations are used in this manuscript:

Abbreviation	Definition
AI	Artificial Intelligence
ML	Machine Learning
DL	Deep Learning
ANN	Artificial Neural Network
RNN	Recurrent Neural Network
LSTM	Long Short-Term Memory Network
Bi-LSTM	Bidirectional Long Short-Term Memory
CNN	Convolutional Neural Network
DCNN	Deep Convolutional Neural Network
MLP	Multilayer Perceptron
SVM	Support Vector Machine
ELM	Extreme Learning Machine
XGBoost	Extreme Gradient Boosting
TFT	Temporal Fusion Transformer
DRL	Deep Reinforcement Learning
DDPG	Deep Deterministic Policy Gradient
GWO-BP	Grey Wolf Optimizer–Backpropagation Neural Network
APSO-Fuzzy-PID	Adaptive Particle Swarm Optimization–Fuzzy–Proportional Integral Derivative Controller
RMPC-AI	Robust Model Predictive Control with Artificial Intelligence
SSA-CNN-LSTM	Salp Swarm Algorithm–Convolutional Neural Network–Long Short-Term Memory
PSO	Particle Swarm Optimization
GBRT	Gradient Boosting Regression Trees
MOORA	Multi-Objective Optimization on the Basis of Ratio Analysis
M-SWARA	Modified Stepwise Weight Assessment Ratio Analysis
TOPSIS	Technique for Order Preference by Similarity to Ideal Solution
Q-SWARA / Q-Theory	Quantum Theory–based Multi-Criteria Decision Model
IoT	Internet of Things
WSN	Wireless Sensor Network
NB-IoT	Narrowband Internet of Things
LoRa	Long Range Communication Technology
CFD	Computational Fluid Dynamics
BIM	Building Information Modeling
FSPM	Functional–Structural Plant Modeling
PV	Photovoltaic
ST	Solar Thermal
PV/T	Photovoltaic–Thermal Hybrid System
CPV/T	Concentrated Photovoltaic–Thermal System
BIPV	Building-Integrated Photovoltaics
TSP	Transparent Solar Panels
TES	Thermal Energy Storage
PCM	Phase Change Material
SHS	Sensible Heat Storage
LHS	Latent Heat Storage
TCES	Thermochemical Energy Storage
GSHP	Ground Source Heat Pump
SOC	State of Charge (Battery)
HVAC	Heating, Ventilation, and Air Conditioning
PAR	Photosynthetically Active Radiation
PFAL	Plant Factory with Artificial Lighting
NZEG	Near Zero Energy Greenhouse
SDGs	Sustainable Development Goals
LCA	Life Cycle Assessment
Agri-PV / Agrivoltaics	Combined Agricultural Production and Photovoltaic Energy Generation
PV-CEA	Photovoltaic-Controlled Environment Agriculture
EWMS	Energy–Water Management System
MPPT	Maximum Power Point Tracking
RMSE	Root Mean Square Error
MAE	Mean Absolute Error
R²	Coefficient of Determination
ANN–SVM	Artificial Neural Network–Support Vector Machine Hybrid
RNA (Modelos RNA)	Artificial Neural Network (Spanish acronym)
PVG	Photovoltaic Generation (when used in cited works)

 

 

 

The Introduction is short and needs revision. It should ideally begin with a broader context leading readers to more specific aspects. Briefly describe all the aspects such as thermal storage, phase change materials, agrovoltaics, …

Reply. We sincerely thank the reviewer for this valuable observation. Following the recommendation, the Introduction section has been substantially expanded and reorganized to provide a broader and more coherent context that gradually guides the reader from global sustainability challenges toward the specific technological aspects addressed in this review.

First, the revised Introduction now begins with a wide contextual framing related to population growth, resource pressure, and the increasing energy demand of protected agriculture. Subsequently, we incorporated a detailed conceptual discussion of thermal energy storage (TES) technologies, including sensible, latent, and thermochemical storage, as well as an explanation of the role of phase change materials (PCM) and nanomaterials in enhancing microclimate stability and energy efficiency. In addition, a dedicated section now highlights recent advances in agrivoltaics as a complementary strategy for solar-assisted greenhouse systems, integrating evidence from empirical and review studies to describe its benefits in microclimate regulation, water-use efficiency, and climate resilience.

Finally, the Introduction integrates the convergence of photovoltaic systems, Internet of Things (IoT), and Artificial Intelligence (AI), outlining how these technologies enable predictive control, energy optimization, and autonomous management in protected agriculture. Overall, the Introduction now provides a complete, logically structured, and well-contextualized framework that covers all the aspects suggested by the reviewer—thermal storage, PCM, agrivoltaics, solar technologies, and AI/IoT integration—thus strengthening the conceptual foundation of the manuscript.

The introduction is organized as follows:

The accelerated growth of the global population, driven by technological progress, has intensified pressure on natural resources and poses significant challenges to the sustainable provision of food, water, and energy (Khan & Pervaiz, 2013; Yildirim & Bilir, 2017). Greenhouses, which play a strategic role in food security, are also intensive consumers of resources and relevant emitters of greenhouse gases (GHG) (Guo et al., 2021), and their long-term viability depends on economic, technological, and policy factors (Wang et al., 2017). The expansion of protected cultivation, stimulated by urbanization and the growing demand for high-quality food, highlights the need for more sustainable agricultural systems. However, the energy consumption of different greenhouse models remains a major challenge (Omidi-Arjenaki et al., 2016), revealing persistent gaps in the understanding of how structural design and automation influence efficiency and sustainability (Ghasemi-Mobtaker et al., 2024). The continued dependence on fossil fuels further increases energy consumption and exacerbates sustainability concerns (Elyasi et al., 2022).

In this context, solar energy in its photovoltaic (PV) and solar thermal (ST) forms emerges as one of the most promising renewable sources to achieve energy autonomy in agricultural greenhouses and dryers. The integration of hybrid solar systems with thermal energy storage (TES) technologies has proven effective in reducing dependence on the electrical grid and improving energy stability under variable climatic conditions (Soussi et al., 2025). Recent studies highlight the relevance of thermal storage particularly sensible, latent, and thermochemical storage to buffer diurnal temperature fluctuations and maintain operational stability in solar-assisted systems. Phase change materials (PCM) provide high energy density and contribute to microclimate stabilization. Evidence shows that PCM enhanced with nanomaterials can increase thermal conductivity and accelerate charging/discharging processes, while nanofluids used as heat-transfer media improve thermal efficiency and storage capacity in hybrid solar systems (Kumar et al., 2024; Verma et al., 2024).

In parallel, agrivoltaics approaches have demonstrated that installing photovoltaic modules over agricultural soils or greenhouse structures can successfully combine electricity generation with crop production. Early studies showed that pairing shade-tolerant crops with PV generation can increase the economic value of the system by up to 30% relative to conventional agriculture, without compromising crop yields (Dinesh & Pearce, 2016). More recent experiments have also demonstrated that partial PV shading reduces leaf temperature, decreases evapotranspiration, and improves water-use efficiency, thereby strengthening the microclimatic advantages of agrivoltaics in arid and semi-arid regions (Barron-Gafford et al., 2019). Additional empirical evidence shows that agrivoltaics systems enhance water conservation, thermal buffering, and climate resilience, with reductions of 20–40% in water consumption and measurable increases in biomass under moderate PV shading  (Barron-Gafford et al., 2019; Hassanpour Adeh et al., 2018). Recent reviews also indicate that agrivoltaics can mitigate heatwaves and frost events, reducing interannual yield variability and positioning this technology as a key strategy for future food security (Widmer et al., 2024).

Against this background, the convergence of solar energy, the Internet of Things (IoT), and Artificial Intelligence (AI) are redefining the concept of intelligent greenhouses. IoT-based monitoring systems allow real-time acquisition of temperature, humidity, radiation, CO₂ concentration, PV generation, and electricity consumption, while machine learning models optimize ventilation, shading, irrigation, and thermal storage (Bicamumakuba et al., 2025; Ghiasi et al., 2023). Recent studies have shown that integrating wireless sensor networks with intelligent control can reduce energy demand and improve water-use efficiency, even in PV-powered greenhouses (Don Chua et al., 2024; Volosciuc et al., 2024). Deep-learning algorithms applied to solar-agricultural systems have demonstrated high performance in PV energy prediction (Dimitriadis et al., 2025), fault detection, MPPT (Maximum Power Point Tracking) optimization, and microclimate control, supporting the development of autonomous, resilient, and low-carbon protected agriculture (Huang et al., 2022; Nautiyal et al., 2025).

However, despite the growing number of studies on AI applied to solar energy, a comprehensive understanding of the field remains fragmented. Most research addresses isolated components such as climate control, energy efficiency, photovoltaic modeling, or autonomous decision-making without articulating a unified conceptual and technological framework. Therefore, a combined bibliometric and systematic review is essential to identify global trends, knowledge gaps, and emerging research directions linking AI, solar energy, and protected agriculture [12]. While bibliometrics quantifies the dynamics of knowledge production, collaboration networks, and thematic structures, the systematic review deepens the analysis by synthesizing methodological and technical advances that reveal how AI, ML, and IoT are reshaping agricultural energy systems toward more efficient, autonomous, and sustainable configurations.

The purpose of this study is to analyze the conceptual and technological evolution of AI applied to solar systems in protected agriculture, and to understand how these tools enhance energy efficiency, automation, and sustainability in controlled agricultural environments. Within this framework, four research questions (RQ) guide the review:

RQ1: How has scientific production and collaboration on AI and solar energy in controlled agriculture evolved during 2012–2025, and what bibliometric patterns explain its consolidation?
RQ2: Which technologies and algorithms such as photovoltaic–thermal (PV/T) systems, IoT-based sensing, and machine learning (ML) have demonstrated the greatest maturity and impact on the energy and climate management of greenhouses and solar dryers?
RQ3: What technical, methodological, and sustainability limitations identified in the literature constrain the adoption of these systems in real production contexts?
RQ4: What future research and development directions emerge from the convergence between AI, renewable energy, and agricultural sustainability toward autonomous, resilient, and low-carbon systems?

These questions reflect an analytical and forward-looking perspective aimed at interpreting the current state of knowledge, identifying critical gaps, and projecting the technological opportunities that will define next-generation intelligent solar agriculture. To address them, this work adopts an integrative approach that combines bibliometric analysis with a systematic review, linking the quantitative evolution of scientific production with the conceptual and technical interpretation of advances in the field. Through the examination of co-authorship networks, keyword co-occurrence, and bibliographic coupling, the main intellectual, thematic, and technological cores shaping global research on AI and solar energy in protected agriculture are identified. In parallel, the synthesis of representative studies highlights key developments such as AI-assisted thermal modeling, predictive microclimate control, energy optimization in PV/T systems, and intelligent maintenance strategies within a sustainability framework. Beyond describing current trends, this review aims to project a strategic vision of the field, demonstrating how the integration of AI and solar energy constitutes a cornerstone for the design of intelligent greenhouses, rural energy transition, and climate adaptation in future agriculture.

 

 

Refer and discuss advances in thermal storage employing different techniques (https://link.springer.com/article/10.1007/s11356-023-31718-8, https://www.sciencedirect.com/science/article/abs/pii/S2451904924003354)

Reply. The suggested documents were analyzed and included in the introduction as a conceptual and referential framework for the systems worked on.

 

Redevelop all the graphs using a professional graphing software.

Reply. We thank the reviewer for this valuable suggestion. In response, all figures have been fully regenerated using professional graphing workflows. Specifically, every plot was reconstructed in R (ggplot2) with improved aesthetics, balanced color scales, and optimized label placement. Additionally, each figure has been exported in high-resolution (900 dpi) PNG and in scalable vector formats (SVG) to ensure maximum clarity and publication-grade quality at any magnification level. These enhanced versions replace all previous figures in the revised manuscript.

 

Section 3.4; Line 311: What is the reason behind more evenly distributed production in Europe?

We thank the reviewer for this valuable observation. In the revised manuscript, we expanded the explanation regarding the more evenly distributed production in Europe. We incorporated recent bibliometric evidence showing that the European Union’s coordinated legislative frameworks—particularly the Renewable Energy Directive and the 2021 regulation on Renewable Energy Communities—have stimulated parallel research development across multiple countries. These shared policy instruments and funding mechanisms promote cross-border collaboration and prevent output from being concentrated in a single nation. This new discussion has been added to Section 3.4 and is now highlighted in the manuscript.

This discussion was modified as follows:

In Europe, production is more evenly distributed, with consistent contributions from the United Kingdom (5), Sweden (3), Spain (3), Italy (3), and Germany (1). This balanced distribution is strongly influenced by the European Union’s coordinated leg-islative and funding frameworks, which promote multi-country participation rather than concentrating scientific output in a single nation. Recent bibliometric analyses show that EU-wide policies, such as the Renewable Energy Directive and the 2021 reg-ulation on Renewable Energy Communities (REC) have stimulated parallel research development across Italy, Austria, Germany, Portugal, and the Netherlands, resulting in diversified but complementary scientific contributions [35–37]. The rapid expansion of REC-related research, with an annual growth rate of 42.82% and more than 200 publi-cations predominantly from EU countries, further demonstrates how shared policy in-struments and coordinated funding mechanisms foster cross-border collaboration and drive a more evenly distributed scientific landscape in Europe. This pattern suggests a more diversified scientific collaboration, where studies focus on modeling, thermal op-timization, and sustainable energy policies. Although smaller in volume than Asia’s output, European participation demonstrates a high level of interdisciplinarity and a strong technical-scientific focus on energy efficiency [38], hybrid system integration [39], and greenhouse monitoring and automation [40].

In the American and African contexts, contributions are still incipient but strate-gically significant. In the Americas, the United States (4), Canada (4), Mexico (3), and Colombia (1) represent focal points for technological transfer toward smart agriculture. In Africa, countries such as Algeria (5), Egypt (3), and South Africa (1) reveal a growing interest in solar energy solutions adapted to arid climatic conditions. Overall, the global distribution shows that research on Artificial Intelligence (AI) and solar energy applied to agriculture is in a consolidation phase, with strong Asian and European leadership and significant potential for expansion in Latin America and Africa, where challenges related to energy sustainability and food security are most critical.

 

Section 3.11: More studies on Building-Integrated Photovoltaics should be discussed as it is a high potential pathway.

We appreciate the reviewer’s valuable recommendation to expand the discussion on Building-Integrated Photovoltaics (BIPV), given its growing relevance and high potential for solar-assisted greenhouse systems. In response, we have substantially revised Section 3.11 by incorporating eight additional peer-reviewed studies, including experimental, modeling, and review papers covering semi-transparent PV, PV/T integration, spectrally selective modules, and next-generation tunable-transmittance or perovskite-based BIPV technologies. These additions provide readers with a more comprehensive technical framework and a clearer understanding of the design principles, microclimatic effects, and agronomic implications of BIPV integration in protected agriculture. The expanded discussion is now included in the revised manuscript and marked for the editor’s convenience.

The section ended up as follows:

From the perspective of Building-Integrated Photovoltaics (BIPV), Liao et al. [34] distinguished fully covered configurations, suitable for low-light-demand crops, and partially covered designs (20–80% shading), appropriate for common vegetables. The study highlighted a thermal penalty of approximately 0.48%/°C in PV efficiency and proposed PV/T (photovoltaic–thermal) integration to cool the modules, stabilize the microclimate, and valorize residual heat for heating or drying. This strategy positions the BIPV–PV/T pathway as a key design axis for self-sufficient greenhouses, where shading patterns and thermal recovery are established as critical control variables. In a parallel review, Yao et al. [60] conducted a comprehensive analysis of the interaction among light, temperature, and humidity, identifying seven priorities for the future of solar-protected agriculture: integration of renewable energies, automated microclimate management, consideration of soil as a thermal modulator, use of selective light spectra, regionalization of radiation models, life cycle assessment, and standardization of design criteria.

Recent experimental and modeling studies conducted in Mediterranean and tem-perate climates further reinforce the potential of BIPV as a high-impact design strategy for solar-assisted agriculture. Evidence from photovoltaic greenhouses shows that semi-transparent or spectrally selective BIPV modules installed on roofs or façade elements can supply a substantial share of the on-site electricity demand often supporting a significant portion of ventilation, heating, and circulation loads while simultaneously reducing solar heat gains and moderating indoor temperature [61–66]. Complementarily, recent reviews highlight that emerging tunable-transmittance PV technologies, organic photovoltaics, and perovskite–silicon tandem structures can optimize the trade-off between PAR transmission and energy generation, enabling envelope designs that adapt dynamically to crop requirements throughout the production cycle [67–70]. Together, these findings underscore that BIPV from conventional crystalline modules to next-generation flexible or semi-transparent devices constitutes one of the most promising pathways for achieving energy-autonomous, climate-resilient greenhouse systems.

 

A table reflecting significant studies on the synergy among IoT, machine learning, and renewable energies for climate-resilient agricultural ecosystems should be added.

Reply. Thank you for this valuable suggestion. Following your recommendation, we have added a new table (Table 1) in Section 3.12, summarizing significant peer-reviewed studies that integrate IoT technologies, Machine Learning algorithms, and renewable-energy systems for climate-resilient agricultural applications. This table enhances the clarity and completeness of Section 3.12 by providing readers with a concise yet comprehensive reference to the most relevant contributions in this rapidly evolving field.

To strengthen the conceptual and empirical understanding of the synergy among IoT architectures, machine learning techniques, and renewable energy systems in controlled agriculture, Table 1 synthesizes representative studies that illustrate concrete implementations of climate-resilient, energy-aware smart greenhouse systems. These works provide evidence of how IoT sensing, ML-based decision models, and photovoltaic energy converge to enhance microclimate regulation, energy autonomy, and operational sustainability.

Table 1. Key Studies Demonstrating Synergy Among IoT, Machine Learning, and Renewable Energy Systems in Controlled Agriculture.

 

Nº	Title (Reference)	Key Contribution (Academic Summary)
1	IoT-based monitoring and control for optimized plant growth in smart greenhouses using soil and hydroponic systems [84].	This study demonstrates the functional integration of IoT sensors, actuators, and a PV system with MPPT, enabling autonomous control of microclimate and water management. The hybrid soil–hydroponic PV-powered greenhouse provides strong empirical evidence of energy self-sufficiency and intelligent climate regulation in protected systems.
2	A comprehensive review on smart and sustainable agriculture using IoT technologies [87].	This systematic review synthesizes the state of the art in the convergence of IoT, machine learning, and renewable energy, emphasizing how these technologies enable predictive, resilient, and low-carbon agricultural systems. It provides a rigorous conceptual foundation for understanding the digital transition toward agro-energetic sustainability.
3	IoT-enabled Greenhouse Systems: Optimizing Plant Growth and Energy Use [88].	The work analyzes IoT architectures for microclimate management using adaptive control algorithms, reporting improvements in energy consumption and stability of critical variables (temperature, humidity, radiation). The study demonstrates that distributed sensing and real-time analytics are essential pillars for energy-optimized protected agriculture.
4	Smart sustainable greenhouses utilizing microcontroller and IOT in the GCC countries; energy requirements y economical analyses study for a concept model in the state of Qatar [89].	This paper experimentally validates that IoT-PV greenhouses can operate with high levels of energy self-sufficiency while maintaining climatic parameters within agronomic ranges. It confirms that solar energy integration, combined with automated control, reduces operational costs and enhances resilience to climatic variability.
5	Environmental monitoring of a smart greenhouse powered by a photovoltaic system [86].	The study evaluates an environmental monitoring system integrated into a fully autonomous PV array, demonstrating stable energy supply and continuous data acquisition. The key contribution lies in validating the compatibility of advanced monitoring with stand-alone photovoltaic operation, particularly relevant in regions with limited infrastructure.

 

Collectively, these studies highlight the growing technological maturity of intelligent, energy-aware greenhouse systems and reinforce the relevance of IoT-enabled architectures integrated with machine-learning algorithms and renewable energy sources. Empirical evidence confirms that these convergent technologies constitute a foundational pathway toward next-generation climate-resilient, near-zero-energy protected agriculture.

 

There is no need for Section 4. Remove it.

Reply.  We appreciate the reviewer’s observation. In accordance with the recommendation, Section 4 has been completely removed in the revised manuscript. Its content was either redundant with the discussion already presented in Section 3 or was redistributed to maintain coherence and improve the logical flow of the manuscript. The updated structure is now more concise and aligned with the reviewer’s suggestion

 

Improve the Conclusions section. The Conclusions section should indicate research gaps and research directions identified as the results of research presented. Explain the significance of your work more clearly and explicitly.

Reply. We sincerely thank the reviewer for this valuable observation. In response, the Conclusions section has been substantially improved to more clearly articulate (i) the research gaps identified through the bibliometric and systematic analyses, (ii) the main research directions emerging from these gaps, and (iii) the significance and contributions of our study.

The new section now is:

The bibliometric analysis confirms a rapid consolidation of the field, with scientific output increasing from 3 publications in 2020 to 27 in 2025, and a total corpus of 79 studies included in this review. Engineering (55 documents), Computer Science (29), and Energy (19) emerged as the dominant subject areas, while China (11 publications) and the United Kingdom (5) led global contributions. Keyword co-occurrence revealed “solar energy” and “solar power generation” as the most frequent terms (10 occurrences each), underscoring the centrality of energy optimization and AI-driven modeling. These indicators demonstrate that AI-assisted solar systems in protected agriculture have reached an important stage of scientific maturity and thematic coherence.

From a systematic perspective, the reviewed studies show that the integration of Artificial Intelligence (AI), Machine Learning (ML), and Internet of Things (IoT) with photovoltaic and thermal technologies enables intelligent greenhouses and solar dryers capable of reducing energy consumption by more than 50%, stabilizing microclimatic conditions, and optimizing water use. Hybrid PV/T systems, predictive control architectures, and phase change materials (PCM) currently constitute the core technological pathways toward near-zero-energy protected agriculture.

However, the synthesis of evidence reveals several research gaps that must be addressed. First, most studies rely on small-scale prototypes or controlled experimental setups, leaving a limited understanding of system performance under real production conditions. Second, there is a lack of standardized metrics for evaluating energy, water, and carbon efficiency across heterogeneous greenhouse designs. Third, the integration of AI models with physical modeling tools such as CFD simulations, digital twins, and multi-energy management frameworks remains incipient, indicating a methodological gap in harmonizing data-driven and physics-based approaches. Finally, social, economic, and policy dimensions of AI solar technologies in agriculture are still underexplored, particularly in regions of the Global South where adoption barriers persist.

Based on these gaps, several research directions emerge. Future work should prioritize (i) large-scale validation of AI–solar systems in commercial greenhouses, (ii) development of unified performance indicators for energy–water–carbon optimization, (iii) integration of CFD, digital twins, and real-time predictive control into unified operational platforms, and (iv) techno-economic and life-cycle assessments that evaluate long-term sustainability and replicability. Additionally, expanding research in Latin America and Africa is essential to ensure equitable access to climate-resilient and low-carbon production systems.

Overall, this review demonstrates that the convergence of AI, ML, IoT, and solar energy is shaping a new generation of autonomous and sustainable agricultural ecosystems. By articulating global trends, technological advances, and future research needs, this work provides a comprehensive framework that supports the design of next-generation agrivoltaic and solar-assisted systems aimed at improving productivity, resource efficiency, and climate resilience in protected agriculture.

 

There are a few grammatical and structural errors. The whole manuscript should be revised to improve the language.

Reply. We thank the reviewer for highlighting the need to improve the grammatical and structural quality of the manuscript. In response, we performed a comprehensive language revision throughout the entire document, including grammar, syntax, academic style, and coherence across sections. The manuscript was fully edited to ensure clarity, precision, and conformity with academic English standards

 

 

Best regards

The authors

Author Response File: Author Response.docx

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

Accept