Comparative Analysis of Machine Learning Models for Accurate Solar Energy Forecasting

To keep the grid stable, to improve storage and to help the energy market work, accurate solar power prediction is needed. The Calgary dataset was used in the study to compare machine learning and deep learning models—XGBoost, CatBoost, LightGBM, Long Short-Term Memory (LSTM) and Temporal Convolutional Networks (TCN)—for short-term energy forecasting. The study also follows a method that includes preprocessing, feature engineering and model training to show the weather and time relationships. The coefficient of determination (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{R}^{\mathbf{2}}$</tex>), the Symmetric Mean Absolute Percentage Error (sMAPE), the Mean Absolute Error (MAE) and the Root Mean Squared Error (RMSE) were used to assess the models. The findings show that gradient boosting techniques regularly beat deep learning models in this setting. XGBoost got the highest accuracy, recording an <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{R}^{\mathbf{2}}$</tex> of 0.997 and an RMSE of 3.64 kWh. It outperforms CatBoost and LightGBM. CatBoost and LightGBM which had higher error rates. Conversely, LSTM and TCN exhibited subpar performance, suggesting challenges in managing feature-sparse datasets with restricted temporal data. The findings indicate that gradient boosting techniques, especially XGBoost, are resilient, effective and ideally suited for enhancing renewable energy integration.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.