Enhancing Solar Energy Production Forecasting with Ensemble-based Learning Techniques

The global trend in renewable energy solutions has emphasized the urgent need for accurate forecasting of solar energy production. This study examines the potential of ensemble-based learning techniques in predicting solar energy production by leveraging historical data collected from 11 solar photovoltaic installations in Calgary, Canada, spanning from September 2015 to March 2023. The dataset was carefully preprocessed to handle missing values, duplicates, and to extract meaningful features that align with the nature of time series data. A range of ML models were implemented and assessed by several forecasting evaluation metrics, including Mean Absolute Error (MAE), Mean Squared Error (MSE), and R-squared <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$(\mathrm{R}^{2})$</tex>. The experimental results demonstrate the superior performance of ensemble methods, particularly XGBoost and Light GBM, in solar energy forecasting. XGBoost achieved the best performance, with an MAE of 25.36, an MSE of 3599.66, and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathrm{R}^{2}$</tex> of 0.957, outperforming models such as Random Forest and Neural Networks. The practical implications of these findings suggest that accurate solar energy forecasting can significantly enhance operational planning and optimize energy management in solar power plants, contributing to more reliable and efficient renewable energy systems.