Wind Power Forecasting Using a New and Robust Hybrid Metaheuristic Approach: A Case Study of Multiple Locations

Wind is a type of renewable energy which has stochastic and intermittent structure. Almost every 10% error made in the wind speed forecast, leads to about 30% error in wind energy generation prediction. Therefore, accurate estimation of the wind speed is vital for wind farms which are increased rapidly. For this aim, a new accurate, quick, and robust hybrid metaheuristic model is used for wind power forecasting. The model is developed based on Radial Movement Optimization (RMO) and Particle Swarm Optimization (PSO). To compare the performance of the designed hybrid metaheuristic model with existing hybrid algorithms, some hybrid metaheuristic models were developed based on the literature, using MATLAB software. The data were obtained from wind measurement stations in two different locations at Burdur and Osmaniye cities, Turkey. Wind power forecasting studies were performed using temperature, humidity and pressure data, except to the wind speed data. Forecasting studies of without using wind speed data, is the most significant factor that challenges the success of all hybrid models. The results show that the error values of ANNs trained with PSO+RMO model were the lowest compared with the other hybrid models.