Aug 2026· Energies· Vol 19, pp. 3904· 0 citations· 26 references
Abstract
Accurate air conditioning (AC) load prediction under extreme weather conditions is critical for power grid stability and energy management. While tree-based ensemble methods such as XGBoost, LightGBM, and Gradient Boosting have become the dominant paradigm in short-term load forecasting, their effectiveness for weather-sensitive AC load prediction—particularly during extreme weather events—remains insufficiently examined. This study presents a systematic comparison of four machine learning models—Support Vector Regression with RBF kernel (SVR-RBF), LightGBM, XGBoost, and Gradient Boosting—for daily AC load estimation conditioned on measured same-day meteorological and calendar features. Based on five years of processed load and weather data from a major city in Southwest China, we construct 31 features and evaluate model performance across four scenarios: normal days, weather-extreme days, high-load P85, and high-load P90. Hyperparameters are selected within the first four years by expanding-window validation, and the fifth year is held out for testing. Our results reveal that SVR-RBF achieves an overall R2 of 0.9772, substantially outperforming LightGBM (0.9091), XGBoost (0.8972), and Gradient Boosting (0.9094); paired moving-block bootstrap intervals for the tree-minus-SVR MAE differences exclude zero. The advantage of SVR-RBF is most pronounced under extreme conditions: on weather-extreme days, SVR-RBF attains R2=0.8316 versus R2=0.39 for the best tree model. This pattern is consistent with the smooth U-shaped temperature–load relationship captured by the RBF kernel. Additional trend-sensitivity analysis shows that annual load growth and target-level extrapolation also explain a substantial part of the tree-model degradation. Furthermore, we quantify the performance limitation of the restricted feature set: while normal-day estimation achieves R2=0.984, high-load P90 days reach R2=0.843. These findings support SVR-RBF as a strong baseline for measured-weather conditional AC load estimation while emphasizing that temporal shift and training-domain coverage must be considered when interpreting model differences. For energy-system applications, the lower errors under high-load and weather-extreme conditions are relevant to peak-demand assessment, reserve planning, and demand-side management.
An integrated probabilistic forecasting framework with three linked stages: extreme-weather load identification, TimeGAN-based sample augmentation, and conformal quantile forecasting, which improves forecasting accuracy under extreme-weather conditions.
Hao Zhang, Xi-Yang Liu, Ruotian Gao et al.· International journal of pat...· 0 citations
Long-term solar energy forecasting accuracy is an essential factor in building sustainable solar infrastructures in rural India. Solar output is highly sensitive to factors like weather fluctuations, seasonal cycles and regional climate, all of which point towards the need for advanced forecasting methods. The study created a hybrid, weather-driven forecasting framework to improve solar energy long-term prediction accuracy through the use of two statistical time-series models ARIMA and SARIMA and two machine learning ML models Random Forests and Long Short-Term Memory LSTM; to account for both linear seasonal trends and complex non-linear relationships in multivariate data such as solar irradiance, temperature, humidity and cloud cover; trained and evaluated on data covering multiple climate types. The hybrid model achieved a prediction accuracy of 92.8% when comparing with commonly used individual models, where ARIMA and SARIMA accuracies ranged between 84.3% - 88.6%; Random Forest was 89.7%, and LSTM was 91.2%. The hybrid model achieved a Mean Absolute Percentage Error MAPE of less than 7.2% and was proven to be more robust and reliable for predicting long-term solar output across varying climates compared to using statistical-only models and proved the benefit of using statistical and ML modelling combined for forecasting long-term solar generation; provided further support to aid rural energy planning, grid integration and policy creation.
Pranay Salikuti, Nikunj Parikh· 2026 International Conferenc...· 0 citations
Long-term precipitation prediction remains a challenging task due to the nonlinear and dynamic interactions among meteorological variables, particularly in regions with complex climatic conditions. This study proposes a Machine Learning framework based on XGBoost combined with temporal feature engineering for predicting rainfall, snowfall, and total precipitation in North Macedonia. Given the limitations of traditional numerical weather prediction models and the lack of region-specific studies focused on the Balkan climate, this research investigates whether advanced supervised learning algorithms can improve the accuracy and stability of precipitation forecasting.A comprehensive dataset covering the period from 2000 to 2025 was obtained from Visual Crossing, followed by extensive preprocessing and feature engineering. The engineered features include lag variables, rolling statistics, seasonal indicators, and interaction terms to better capture temporal dependencies and complex relationships in the data.Three predictive models were developed: a rainfall classifier, a snowfall classifier, and a general precipitation classifier. The primary algorithm employed was XGBoost, while the snowfall prediction model utilized an ensemble stacking approach combining XGBoost, Random Forest, and LightGBM to address class imbalance and enhance model robustness. Model validation was conducted using a TimeSeriesSplit strategy with temporal gaps to prevent data leakage.The results demonstrate significant performance improvements using gradient boosting-based methods. The snowfall model achieved an accuracy ranging from 90% to 97%, rainfall prediction reached 81% to 84%, and the general precipitation model achieved 80% to 81% accuracy. Feature importance analysis identified temperature, humidity, atmospheric pressure patterns, and cloud cover as the most influential predictors. These findings confirm that machine learning provides a reliable and effective approach for long-term precipitation forecasting in the region, with potential applications in agriculture, transportation, and environmental planning.
Nexhmije Miftari, Valon Ahmeti, Majlinda Axhiu· The eurasia proceedings of s...· 0 citations
Accurate short‐ and medium‐term solar irradiance forecasting is vital for integrating solar power into the grid, but high variability and weather dependence make it challenging. Machine learning (ML) models offer promise, but their performance often depends on forecast horizon and training data size. In this study, four models, including random forest (RF), support vector regression (SVR), gradient boosting regressor (GBR), and deep neural networks (DNN), are compared for 15‐min, hourly, and day‐ahead solar irradiance prediction. The models are trained with meteorological and irradiance data for the city of Sioux Falls, South Dakota, using datasets ranging from 1 to 8 years (2012–2019). A total of 96 models are developed by combining four algorithms across three horizons and eight training data sizes. Using the maximum relevance minimum redundancy (MRMR) technique, the best features are selected, and the results are evaluated using standard error metrics. The results indicate that RF provides the most stable accuracy prediction, SVR provides the best accuracy on day‐ahead forecasts with limited data, GBR performs best at the hourly prediction level, and DNN accuracy is more sensitive to both data quality and overfitting. It emphasizes the importance of choosing models that align with the forecast horizon and the size of available data. In this paper, we provide an in‐depth comparison of four state‐of‐the‐art ML models across various horizons and training data sizes. Results show the importance of integrating MRMR, with guidelines for researchers and energy planners to achieve more accurate and efficient solar irradiance prediction.
Fatemeh Keramati, H. Mohammadi· Journal of Forecasting· 0 citations
Weather prediction remains a critical challenge due to the nonlinear and dynamic nature of atmospheric systems, as traditional numerical weather prediction (NWP) models struggle to process large, high-dimensional meteorological data and often lack the adaptability needed for accurate short- and medium-term forecasts, particularly during extreme weather events. This study develops an improved AI-driven weather prediction system that enhances forecasting accuracy for temperature, humidity, wind speed, and atmospheric pressure through a Random Forest predictive model integrated with the Open Weather Map API and geolocation services. Using Agile methodology, data was collected, preprocessed, and trained in Python, while a web-based interface was built with JavaScript/TypeScript and React for visualization. The proposed system achieved approximately 87% short-term forecasting accuracy (87% applies to 1 – 7-day short term forecast), demonstrating improved precision and enhanced early-warning capability for extreme events. Comparative evaluation showed that, whereas the existing edge-based system was constrained by low processing power, maintenance overhead, and security concerns, the proposed AI system outperformed it in accuracy, adaptability, and real-time usability, offering significant benefits for agriculture, disaster management, and urban planning.
Idayana Alabere· International journal of res...· 0 citations
The increasing penetration of wind energy and the rapid growth of computing-intensive data centers have made the joint management of renewable-generation uncertainty and flexible electricity demand a pressing concern, particularly during extreme weather events such as storms, cold snaps, and rapid wind ramps. This paper proposes an integrated framework that couples a deep learning-based wind power forecasting model with a data center load response optimization scheme. A convolutional neural network-bidirectional long short-term memory network with an attention mechanism (CNN-BiLSTM-Attention) is developed to capture both local ramp features and long-range temporal dependencies in wind power time series, and an auxiliary extreme-weather risk classifier flags high-uncertainty forecasting windows. The resulting probabilistic forecasts feed a mixed-integer linear programming (MILP) optimizer that reschedules deferrable data center workloads, migrates jobs across sites, and adjusts cooling set-points to align computing load with available wind generation. Simulation results using SCADA wind-farm data and a representative multi-site data center workload show that the proposed forecasting model reduces RMSE by up to 20.7% relative to Transformer and CNN-LSTM baselines under extreme-weather subsets, while the load-response strategy achieves up to 18.7% energy-cost savings with service-level-agreement (SLA) completion above 94% during severe storm scenarios.
Dalin Jiang, Dazheng Liu, Bo Bai et al.· 2026 5th International Confe...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
