Enhancing photovoltaic system performance using advanced gradient boosting methods

Photovoltaic (PV) systems are indispensable elements in clean energy production. Predicting PV power is crucial for optimizing system performance and ensuring grid stability. This paper investigates the performance of gradient boosting approaches, XGBoost and Catboost, in predicting PV power. Datasets recorded each minute, including weather data and power output from two distinct PV systems in Brisbane, Australia, are used in this study for evaluation. Results show that the XGBoost slightly outperformed Catboost and achieved good prediction performance. Also, the XGBoost output was interpreted using the Shapley additive interpretation (SHAP) method to identify the important factor impacting PV power prediction. Results confirm that solar radiation is the most important factor influencing the performance of photovoltaic systems, and the other weather variables have a weaker impact. These findings contribute valuable insights for enhancing the efficiency and reliability of PV systems in renewable energy applications.