Long-Term Wind Speed Projection Based on Machine Learning Regression Techniques in the Perspective of Bangladesh

Abstract

Wind speed projection is a research hotspot in wind energy conversion systems because it aids to optimize the operating costs as well as boost the reliability of power generation from wind. Wind power output depends on wind speed that depends on different parameters. Non-linearity among these parameters makes machine learning methods a preferable approach. In our work, we have used eight parameters and fifteen different machine learning regression methods to predict the hourly wind speed of five different sites in Bangladesh. The results obtained from these methods are very compelling as it has a low Mean Absolute Error (MAE) and Root Mean Square Error (RMSE). So, this sort of investigation can be effective for future wind energy-related ventures and research in Bangladesh.