Optimizing Solar Panel Capacity Forecasting: A Comprehensive Analysis of Ensemble Learning Models in the Saudi Arabian Context

Le résumé fourni par la source

This study investigates the application of ensemble learning models to enhance solar panel capacity forecasting in Saudi Arabia, a nation at the forefront of renewable energy innovation. Focusing on the prediction of solar panel efficiency parameters-maximum power (Pmax), maximum voltage (Vmax), and maximum current (Imax)-the research compares three ensemble methods: Bagging, Additive Regression (boosting), and Stacking. Data was collected from crystalline solar panels in Dharan, Saudi Arabia, using a data logger that recorded environmental variables at 20-second intervals. The models were evaluated using to-fold cross-validation, with Bagging demonstrating superior performance, achieving correlation coefficients (CC) of 0.9748 for Umax, 0.9928 for Pmax, and 0.9923 for Imax, and correspondingly low Root Mean Squared Error (RMSE) values. Additive Regression showed moderate effectiveness, while Stacking struggled with negative CC values and higher RMSE scores, indicating a potential overfitting issue. The findings underscore the importance of selecting appropriate machine learning models for accurate solar energy predictions, with Bagging emerging as the most reliable method in this context.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.