Wind power prediction using Machine Learning models: A comparative study

Abstract

This study addresses the challenge of integrating wind energy into power systems due to its natural variability by employing machine learning techniques to predict wind power generation using real field data from a Senvion MM82 wind turbine in France. Five algorithms—Random Forest, Decision Tree, XGBoost, CatBoost, and Linear Regression—were evaluated using standard performance metrics, with Linear Regression emerging as the most accurate. The research highlights the critical role of data preprocessing, feature engineering, and the inclusion of meteorological and operational parameters, ultimately demonstrating the potential of machine learning to enhance forecasting accuracy, support grid integration, and improve the reliability of renewable energy systems.