Optimizing photovoltaic performance: Data-driven maximum power point prediction via advanced regression models

Abstract

The accurate prediction of the Maximum Power Point (PMPP) in photovoltaic (PV) systems is critical for optimizing energy yield and enhancing solar energy harvesting efficiency. This study explores the application of data-driven methods to improve PMPP prediction, utilizing advanced regression techniques such as Ridge Regression, Lasso Regression, Decision Tree Regression, and Random Forest Regression. By analyzing a dataset of irradiance, temperature, and PMPP measurements, the research compares the performance of these models in capturing complex nonlinear relationships between key variables. Results indicate that tree-based models, particularly Random Forest Regression, outperform linear models, demonstrating superior predictive accuracy and robustness. Feature importance analysis further highlights the dominant influence of irradiance (GPOA) on PMPP, emphasizing the value of precise irradiance data. These findings underscore the potential of machine learning techniques in optimizing PV system performance. Future research should focus on integrating additional features, such as atmospheric conditions and panel characteristics, and exploring deep learning methods to enhance prediction accuracy further.