Wind Energy Forecasting: A Comparative Analysis of Machine Learning Approaches

This paper conducts a comprehensive comparative analysis of machine learning and deep learning techniques for wind energy forecasting. Wind energy prediction is a crucial aspect of renewable energy management. We study and assess the application of tree-based algorithms (LightGBM, CatBoost, and Random Forest) and deep learning approaches (feedforward networks with skip connection and FT-transformers). In the numerical study, we evaluate forecasting accuracy, robustness, computational efficiency, and analyze the reasons why some methods are better than others. This research contributes to advancing wind energy forecasting methodologies in the renewable energy sector. Accurate predictive models play the most crucial role in optimizing wind energy integration into the grid, supporting a sustainable energy future.