Search Results (85)

Environmental comfort and air pollution are among the most important indicators for assessing the population’s quality of life in urban agglomerations. This study aims to explore long-term memory in air pollution time series by analyzing the dynamics of the Hurst exponent and evaluating the predictability index. This type of statistical pre-forecast analysis is essential for developing accurate forecasting models for such time series. The effectiveness of air quality monitoring systems largely depends on the precision of these forecasts. The Ekibastuz coal-mining center, which houses one of the largest coal-fired power stations in Kazakhstan and the world, with a capacity of about 4000 MW, was chosen as an example for the study. Data for the period from 1 March 2023 to 31 December 2024 were collected and analyzed at the Ekibastuz coal-fired power station. During the specified period, 14 indicators (67,527 observations) were collected at 10 min intervals, including mass concentrations of CO, NO, NO₂, SO₂, PM_2.5, and PM₁₀, as well as current mass consumption of CO, NO, NO₂, SO₂, dust, and NO_x. The detrended fluctuation analysis of a time series of air pollution indicators was used to calculate the Hurst exponent and identify long-term memory. Changes in the Hurst exponent in regards to dynamics were also investigated, and a predictability index was calculated to monitor emissions of pollutants in the air. Long-term memory is recorded in the structure of all the time series of air pollution indicators. Dynamic analysis of the Hurst exponent confirmed persistent time series characteristics, with an average Hurst exponent of about 0.7. Identifying the time series plots for which the Hurst exponent is falling (analysis of the indicator of dynamics), along with the predictability index, is a sign of an increase in the influence of random factors on the time series. This is a sign of changes in the dynamics of the pollutant release concentrations and may indicate possible excess emissions that need to be controlled. Calculating the dynamic changes in the Hurst exponent for the emission time series made it possible to identify two distinct clusters corresponding to periods of persistence and randomness in the operation of the coal-fired power station. The study shows that evaluating the predictability index helps fine-tune the parameters of time series forecasting models, which is crucial for developing reliable air pollution monitoring systems. The results obtained in this study allow us to conclude that the method of trended fluctuation analysis can be the basis for creating an indicator of the level of air pollution, which allows us to quickly respond to possible deviations from the established standards. Environmental services can use the results to build reliable monitoring systems for air pollution from coal combustion emissions, especially near populated areas. Full article

Despite notable advancements in smart home technologies, residential energy management continues to face critical challenges. These include the complex integration of intermittent renewable energy sources, issues related to data latency, interoperability, and standardization across diverse systems, the inflexibility of centralized control architectures in dynamic environments, and the difficulty of accurately modeling and influencing occupant behavior. To address these challenges, this study proposes an intelligent multi-agent system designed to accurately estimate and control energy consumption in residential buildings, with the overarching objective of optimizing energy usage while maintaining occupant comfort and satisfaction. The methodological approach employed is a hybrid framework, integrating multi-agent system architecture with system dynamics modeling and agent-based modeling. This integration enables decentralized and intelligent control while simultaneously simulating physical processes such as heat exchange, insulation performance, and energy consumption, alongside behavioral interactions and real-time adaptive responses. The system is tested under varying conditions, including changes in building insulation quality and external temperature profiles, to assess its capability for accurate control and estimation of energy use. The proposed tool offers significant added value by supporting real-time responsiveness, behavioral adaptability, and decentralized coordination. It serves as a risk-free simulation platform to test energy-saving strategies, evaluate cost-effective insulation configurations, and fine-tune thermostat settings without incurring additional cost or real-world disruption. The high fidelity and predictive accuracy of the system have important implications for policymakers, building designers, and homeowners, offering a practical foundation for informed decision making and the promotion of sustainable residential energy practices. Full article

Gas turbine (GT) modeling and optimization have been widely studied at the design level but still lacks focus on real-world operational cases. The concept of a digital twin (DT) allows for the interaction between operation data and the system dynamic performance. Among many DT studies, only a few focus on GT for thermal power plants. This study proposes a digital twin prototype framework including the following modules: process modeling, parameter estimation, and performance optimization. Provided with real-world power plant operational data, key performance parameters such as turbine inlet temperature (TIT) and specific fuel consumption (SFC) were initially unavailable, therefore necessitating further calculation using thermodynamic analysis. These parameters are then used as a target label for developing artificial neural networks (ANNs). Three ANN models with different structures are developed to predict TIT, SFC, and turbine power output (GTPO), achieving high R² scores of 94.03%, 82.27%, and 97.59%, respectively. Physics-informed neural networks (PINNs) are then employed to estimate the values of the air–fuel ratio and combustion efficiency for each time index. The PINN-based estimation resulted in estimated values that align with the literature. Subsequently, an unconventional method of detecting alarms by using conformal prediction were also proposed, resulting in a significantly reduced number of alarms. The developed ANNs are then combined with particle swarm optimization (PSO) to carry out performance optimization in real time. GTPO and SFC are selected as the primary metrics for the optimization, with controllable parameters such as AFR and a fine-tuned inlet guide vane position. The results demonstrated that GTPO could be optimized with the application of conformal prediction when the true GTPO is detected to be higher than the upper range of GTPO obtained from the ANN model with a conformal prediction of a 95% confidence level. Multiple PSO variants were also compared and benchmarked to ensure an enhanced performance. The proposed PSO in this study has a lower mean loss compared to GEP. Furthermore, PSO has a lower computational cost compared to RS for hyperparameter tuning, as shown in this study. Ultimately, the proposed methods aim to enhance GT operations via a data-driven digital twin concept combination of deep learning and optimization algorithms. Full article

This study explores the application of machine learning (ML) techniques—gradient boosting (GB), ridge regression (RR), and support vector regression (SVR)—for estimating the consumption of energy (CE) and Blaine fineness (BF) in cement clinker grinding. This study utilizes key clinker grinding parameters, such as maximum ball size, ball filling ratio, clinker mass, rotation speed, and number of revolutions, as input features. Through comprehensive preprocessing, feature selection methods (mutual info regression (MIR), lasso regression (LR), and sequential backward selection (SBS)) were employed to identify the most significant variables for predicting CE and BF. The performance of the models was optimized using a grid search for hyperparameter tuning and validated using k-fold cross-validation (k = 10). The results show that all ML methods effectively estimated the target parameters, with SVR demonstrating superior accuracy in both CE and BF predictions, as evidenced by its higher R² and lower error metrics (MAE, MAPE, and RMSE). This research highlights the potential of ML models in optimizing cement grinding processes, offering a novel approach to parameter estimation that can reduce experimental effort and enhance production efficiency. The findings underscore the advantages of SVR, making it the most reliable method for predicting energy consumption and Blaine fineness in clinker grinding. Full article

Accurate path loss prediction is essential for optimizing Long-Range Wide-Area Network (LoRaWAN) performance. Previous studies have employed various Machine Learning (ML) models for path loss prediction. However, environmental factors such as temperature, humidity, barometric pressure, and particulate matter have been largely neglected. This study bridges this gap by evaluating the performance of five boosting ML models—AdaBoost, XGBoost, LightGBM, GentleBoost, and LogitBoost—under dynamic environmental conditions. The models were compared with theoretical models (Log-Distance and Okumura-Hata) and existing studies that employed the same dataset based on metrics such as RMSE, MAE, and R². Furthermore, a detailed performance vs. complexity analysis was conducted using metrics such as training time, inference latency, model size, and energy consumption. Notably, barometric pressure emerged as the most influential environmental factor affecting path loss across all models. Bayesian Optimization was applied to fine-tune hyperparameters to improve model accuracy. Results showed that LightGBM outperformed other models with the lowest RMSE of 0.5166 and the highest R² of 0.7151. LightGBM also offered the best trade-off between accuracy and computational efficiency. The findings show that boosting algorithms, particularly LightGBM, are highly effective for path loss prediction in LoRaWANs. Full article

This study proposes an automatic gear shift classification algorithm in M1 category vehicles using data acquired through the onboard diagnostic system (OBD II) and GPS. The proposed approach is based on the analysis of identification parameters (PIDs), such as manifold absolute pressure (MAP), revolutions per minute (RPM), vehicle speed (VSS), torque, power, stall times, and longitudinal dynamics, to determine the efficiency and behavior of the vehicle in each of its gears. In addition, the unsupervised K-means algorithm was implemented to analyze vehicle gear changes, identify driving patterns, and segment the data into meaningful groups. Machine learning techniques, including K-Nearest Neighbors (KNN), decision trees, logistic regression, and Support Vector Machines (SVMs), were employed to classify gear shifts accurately. After a thorough evaluation, the KNN (Fine KNN) model proved to be the most effective, achieving an accuracy of 99.7%, an error rate of 0.3%, a precision of 99.8%, a recall of 99.7%, and an F1-score of 99.8%, outperforming other models in terms of accuracy, robustness, and balance between metrics. A multiple linear regression model was developed to estimate instantaneous fuel consumption (in L/100 km) using the gear predicted by the KNN algorithm and other relevant variables. The model, built on over 66,000 valid observations, achieved an R² of 0.897 and a root mean square error (RMSE) of 2.06, indicating a strong fit. Results showed that higher gears (3, 4, and 5) are associated with lower fuel consumption. In contrast, a neutral gear presented the highest levels of consumption and variability, especially during prolonged idling periods in heavy traffic conditions. In future work, we propose integrating this algorithm into driver assistance systems (ADAS) and exploring its applicability in autonomous vehicles to enhance real-time decision making. Such integration could optimize gear shift timing based on dynamic factors like road conditions, traffic density, and driver behavior, ultimately contributing to improved fuel efficiency and overall vehicle performance. Full article

Predicting the motion of handwritten digits in video sequences is challenging due to complex spatiotemporal dependencies, variable writing styles, and the need to preserve fine-grained visual details—all of which are essential for real-time handwriting recognition and digital learning applications. In this context, our study aims to develop a robust predictive framework that can accurately forecast digit trajectories while preserving structural integrity. To address these challenges, we propose a novel video prediction architecture integrating ConvCARU with a modified DCGAN to effectively separate the background from the foreground. This ensures the enhanced extraction and preservation of spatial and temporal features through convolution-based gating and adaptive fusion mechanisms. Based on extensive experiments conducted on the MNIST dataset, which comprises 70 K pixel images, our approach achieves an SSIM of 0.901 and a PSNR of 29.31 dB. This reflects a statistically significant improvement in PSNR of +0.20 dB (p < 0.05) compared to current state-of-the-art models, thus demonstrating its superior capability in maintaining consistent structural fidelity in predicted video frames. Furthermore, our framework performs better in terms of computational efficiency, with lower memory consumption compared to most other approaches. This underscores its practicality for deployment in real-time, resource-constrained applications. These promising results consequently validate the effectiveness of our integrated ConvCARU–DCGAN approach in capturing fine-grained spatiotemporal dependencies, positioning it as a compelling solution for enhancing video-based handwriting recognition and sequence forecasting. This paves the way for its adoption in diverse applications requiring high-resolution, efficient motion prediction. Full article

This paper presents a Virtualized Computational Radio Frequency Identification (VCRFID) solution that utilizes far-field UHF RF for sensing, computing, and self-powering at the edge. A standard UHF RFID system is asymmetric as it consists of a relatively large, complex “reader”, which acts as an RF transmitter and controller for a number of small simple battery-less “tags”, which work in passive mode as they communicate and harvest RF energy from the reader. Previously proposed Computational RFID (CRFID) solutions enhance the standard RFID tags with microcontrollers and sensors in order to gain enhanced functionality, but they end up requiring a relatively high level of power, and thus ultimately reduced range, which limits their use for many Internet-of-Things (IoT) application scenarios. Our VCRFID solution instead keeps the functionality of the tags minimalistic by only providing a sensor interface to be able to capture desired environmental data (temperature, humidity, vibration, etc.), and then transmit it to the RFID reader, which then performs all the computational load usually carried out by a microcontroller on the tag in prior work. This virtualization of functions enables the design of a circuit without a microcontroller, providing greater flexibility and allowing for wireless reconfiguration of tag functions over RF for a 97% reduction in energy consumption compared to prior energy-harvesting RFID tags with microcontrollers. The target application is Industry 4.0 where our VCRFID solution enables battery-less fine-grain monitoring of vibration and temperature data for pumps and motors for predictive maintenance scenarios. Full article

This study addresses the sentiment classification of short texts in teaching evaluations. To mitigate concerns regarding data security in cloud-based sentiment analysis and to overcome the limited feature extraction capacity of traditional deep-learning methods, we propose a distributed symmetric approximate homomorphic hybrid sentiment classification model, denoted BHE+ALBERT-Mixplus. To enable homomorphic encryption of non-polynomial functions within the ALBERT-Mixplus architecture—a mixing-and-enhancement variant of ALBERT—we introduce the BHE (BERT-based Homomorphic Encryption) algorithm. The BHE establishes a distributed symmetric approximation workflow, constructing a cloud–user symmetric encryption framework. Within this framework, simplified computations and mathematical approximations are applied to handle non-polynomial operations (e.g., GELU, Softmax, and LayerNorm) under the CKKS homomorphic-encryption scheme. Consequently, the ALBERT-Mixplus model can securely perform classification on encrypted data without compromising utility. To improve feature extraction and enhance prediction accuracy in sentiment classification, ALBERT-Mixplus incorporates two core components: 1. A meta-information extraction layer, employing a lightweight pre-trained ALBERT model to capture extensive general semantic knowledge and thereby bolster robustness to noise. 2. A hybrid feature-extraction layer, which fuses a bidirectional gated recurrent unit (BiGRU) with a multi-scale convolutional neural network (MCNN) to capture both global contextual dependencies and fine-grained local semantic features across multiple scales. Together, these layers enrich the model’s deep feature representations. Experimental results on the TAD-2023 and SST-2 datasets demonstrate that BHE+ALBERT-Mixplus achieves competitive improvements in key evaluation metrics compared to mainstream models, despite a slight increase in computational overhead. The proposed framework enables secure analysis of diverse student feedback while preserving data privacy. This allows marginalized student groups to benefit equally from AI-driven insights, thereby embodying the principles of educational equity and inclusive education. Moreover, through its innovative distributed encryption workflow, the model enhances computational efficiency while promoting environmental sustainability by reducing energy consumption and optimizing resource allocation. Full article

Artificial intelligence of things (AIoT) has become a pivotal enabler of precision agriculture by supporting real-time, data-driven decision-making at the edge. Deep learning (DL) models are central to this paradigm, offering powerful capabilities for analyzing environmental and climatic data in a range of agricultural applications. However, deploying these models on edge devices remains challenging due to constraints in memory, computation, and energy. Existing model compression techniques predominantly target large-scale 2D architectures, with limited attention to one-dimensional (1D) models such as gated recurrent units (GRUs), which are commonly employed for processing sequential sensor data. To address this gap, we propose a novel three-stage coarse-to-fine compression framework, termed SUQ-3 (Structured, Unstructured Pruning, and Quantization), designed to optimize 1D DL models for efficient edge deployment in AIoT applications. The SUQ-3 framework sequentially integrates (1) structured pruning with an $M\times N$ sparsity pattern to induce hardware-friendly, coarse-grained sparsity; (2) unstructured pruning to eliminate low-magnitude weights for fine-grained compression; and (3) quantization, applied post quantization-aware training (QAT), to support low-precision inference with minimal accuracy loss. We validate the proposed SUQ-3 by compressing a GRU-based crop recommendation model trained on environmental and climatic data from an agricultural dataset. Experimental results show a model size reduction of approximately 85% and an 80% improvement in inference latency while preserving high predictive accuracy (F1 score: 0.97 vs. baseline: 0.9837). Notably, when deployed on a mobile edge device using TensorFlow Lite, the SUQ-3 model achieved an estimated energy consumption of 1.18 μJ per inference, representing a 74.4% reduction compared with the baseline and demonstrating its potential for sustainable low-power AI deployment in agricultural environments. Although demonstrated in an agricultural AIoT use case, the generality and modularity of SUQ-3 make it applicable to a broad range of DL models across domains requiring efficient edge intelligence. Full article

Heating, ventilation, and air conditioning (HVAC) systems account for 60% of the energy consumption in commercial buildings. Each year, millions of dollars are spent on electricity bills by commercial building operators. To address this energy consumption challenge, a predictive model named Bayesian optimisation Convolution Neural Network Multivariate Long Short-term Memory (BO CNN-M-LSTM) is introduced in this research. The proposed model is designed to perform load forecasting, optimizing energy usage in commercial buildings. The CNN block extracts local features, whereas the M-LSTM captures temporal dependencies. The hyperparameter fine tuning framework applied Bayesian optimization to enhance output prediction by modifying model properties with data characteristics. Moreover, to improve occupant well-being in commercial buildings, the thermal comfort adaptive model developed by de Dear and Brager was applied to ambient temperature in the preprocessing stage. As a result, across all four datasets, the BO CNN-M-LSTM consistently outperformed other models, achieving an 8% improvement in mean percentage absolute error (MAPE), 2% in normalized root mean square error (NRMSE), and 2% in R² score.This indicates the consistent performance of BO CNN-M-LSTM under varying environmental factors, highlight the model robustness and adaptability. Hence, the BO CNN-M-LSTM model is a highly effective predictive load forecasting tool for commercial building HVAC systems. Full article

Accurate building energy consumption prediction is essential for efficient energy management and energy optimization. This study utilizes bidirectional long short-term memory (BiLSTM) to automatically extract deep time series features. The nonlinear fitting and high-precision prediction capabilities of Random Forest (RF) and XGBoost models are then utilized to develop a BiLSTM-RF-XGBoost stacked hybrid model. To enhance model generalization and reduce overfitting, a Bayesian algorithm with an early stopping mechanism is utilized to fine-tune hyperparameters, and strict K-fold time series cross-validation (TSCV) is implemented for performance evaluation. The hybrid model achieves a high TSCV average R² value of 0.989 during cross-validation. When evaluated on an independent test set, it yields a mean square error (MSE) of 0.00003, a root mean square error (RMSE) of 0.00548, a mean absolute error (MAE) of 0.00130, and a mean absolute percentage error (MAPE) of 0.26%. These values are significantly lower than those of comparison models, indicating a significant improvement in predictive performance. The study offers insights into the internal decision-making of the model through SHAP (SHapley Additive exPlanations) feature significance analysis, revealing the key roles of temperature and power lag features, and validating that the stacked model effectively utilizes the outputs of base models as meta-features. This study makes contributions by proposing a novel hybrid model trained with Bayesian optimization, analyzing the influence of various feature factors, and providing innovative technological solutions for building energy consumption prediction. It also provides theoretical value and guidance for low-carbon building energy management and application. Full article

Effective building energy prediction is essential for optimizing energy management, but existing models struggle with data scarcity and sensor heterogeneity across different buildings. Conventional approaches, including centralized and transfer learning methods, fail to generalize well due to varying sensor configurations and inconsistent data availability. To overcome these challenges, this study proposes a Personalized Federated Learning (pFL) framework that integrates multi-level feature masking, model ensemble techniques, and knowledge transfer to enhance predictive performance across diverse buildings. The proposed feature masking strategy extracts the most relevant time-series features, while model ensemble learning improves generalization, and knowledge transfer enables adaptive fine-tuning for each building. These techniques allow pFL to retain global knowledge while personalizing to local energy consumption patterns, making it more effective than traditional FL methods. Experiments conducted on a campus energy dataset demonstrate that pFL consistently outperforms FedAvg, FedProx, and standalone models in energy prediction accuracy. The most significant improvements are observed in buildings with highly fluctuating consumption patterns, validating the effectiveness of the proposed approach in handling heterogeneous sensing environments. This study highlights the potential of Federated Learning for scalable and adaptive energy prediction. Future work will focus on refining multi-horizon forecasting and developing strategies to enhance knowledge sharing among buildings for improved long-term performance. Full article

The global push for sustainable production, driven by initiatives like the Paris Agreement and the European Green Deal, necessitates reducing CO₂ emissions in industrial processes. Additive manufacturing (AM), with its potential for material efficiency and decentralization, offers promising opportunities for lowering carbon footprints. Due to the significant importance of enhancing the performance of AM via the fine-tuning of printing parameters, this study investigates the dual objectives of understanding parameter influences and leveraging artificial intelligence (AI) to predict CO₂ emissions in fused deposition modeling (FDM) processes. A full-factorial experimental design with 81 test prints was conducted, varying four key parameters—layer height, infill density, perimeters, and nozzle temperature—at three levels (min, mid, and max). The results highlight infill density as the most influential factor, significantly impacting material usage, energy consumption, and overall CO₂ emissions. Five AI algorithms were employed for predictive modeling, with XGBoost demonstrating the highest accuracy in forecasting emissions. By systematically analyzing process interdependencies and providing quantitative insights, this study advances sustainable 3D printing practices. The findings offer practical implications for optimizing AM processes, benefiting both researchers and industrial stakeholders aiming to reduce CO₂ emissions without compromising product integrity. Full article

Analyzing the electric power consumption (EPC) patterns of China’s mega urban agglomerations is crucial for promoting sustainable development both domestically and globally. Utilizing 2017–2021 NPP/VIIRS low-light remote sensing imagery to extract total nighttime light data, this study proposed an EPC prediction method based on the K-Means clustering algorithm combined with multiple indicators integrated with socio-economic factors. Combining IPAT theory, regional GDP and population density, the final EPC prediction models were developed. Using these models, the EPC distributions for Beijing-Tianjin-Hebei (BTH), Yangtze River Delta (YRD), and Pearl River Delta (PRD) urban agglomerations in 2017–2021 were generated at both the administrative district level and the 1 km × 1 km grid scale. The spatio-temporal dynamics of the EPC distribution in these urban agglomerations during this period were then analyzed, followed by EPC predictions for 2022. The models showed a significant improvement in prediction accuracy, with the average MARE decreasing from 30.52% to 7.60%, 25.61% to 11.08% and 18.24% to 12.85% for the three urban agglomerations, respectively; EPC clusters were identified in these areas, mainly concentrated in Langfang and Chengde, Shanghai and Suzhou, and Dongguan; from 2017 to 2021, the EPC values of the three urban agglomerations show a growth trend and the distribution patterns were consistent with their economic development and population density; the R² values and the statistical values for the 2022 EPC predictions using the improved classification EPC models reached 0.9692, 0.9903 and 0.9677, respectively, confirming that the proposed method can effectively predict the EPC of urban agglomerations and is applicable in various scenarios. This method provides a timely and accurate spatial update of EPC dynamics, offering fine-scale characterization of urban EPC patterns using night light images. Full article