Enhancing wind energy forecasting in India: A site-specific comparative analysis of machine learning models and Weibull statistics using long-term data

The global pursuit of clean and sustainable energy has elevated wind power as a vital renewable resource. Yet, its inherent variability poses significant challenges for integration into smart grids, where precise forecasting is crucial for grid stability and efficient energy management. Traditional statistical methods often fail to capture the nonlinear dynamics of wind, especially across diverse geographies. This study explores the application of advanced machine learning (ML) techniques to improve hourly wind speed prediction across six Indian locations, using long-term historical data (1969–2006) from the Indian Meteorological Department. Four ML models, Gradient Boosting Machine (GBM), Long Short-Term Memory (LSTM), Convolutional Neural Network (CNN), and Random Forest (RF), are employed to forecast wind power density (WPD), with performance assessed via Root Mean Square Error (RMSE) and Mean Absolute Error (MAE). A novel aspect of this work is the combined use of Weibull statistical modeling and ML forecasting to enhance wind resource assessments at varying hub heights. Results show that ML models, particularly GBM, outperform traditional approaches, yielding higher R2 values, while CNN consistently underperforms. Among the studied sites, Tuticorin records the highest WPD (1090.06 W/m2 in July), and Jaipur the lowest (14.03 W/m2 in December), underscoring regional variability in wind energy potential. This research highlights the importance of selecting appropriate, location-specific models for reliable wind forecasting and offers critical insights for optimizing wind energy planning and integration into smart grid systems.