Data-Driven Wind Speed Forecasting Techniques Using Hybrid Neural Network Methods

Wind power generation is a dominant form of renewable energy sources with significant technical progress over the past decades. One of the major challenges in wind power generation is the intermittent nature of wind speed. In this paper, wind speed prediction techniques are investigated using wind speed measurement data in Saskatchewan, Canada. Despite excellent wind power potential in Saskatchewan, currently, only 6.5 % of total electricity demand is supplied by wind power in this province. In this paper, three hybrid Neural Network methods (Wavelet Neural Network (WNN) trained by Improved Clonal Selection Algorithm (ICSA); WNN trained by Particle Swarm Optimization (PSO); and Extreme Learning Machine (ELM)-based Neural Network) are implemented and compared for wind speed forecasting using actual recorded wind speed data of Saskatchewan, which paves the way for economical operation, planning, and optimization of the current and future wind farms in Saskatchewan.