Wind speed prediction based on genetic neural network

Abstract

Short-term prediction of wind speed is important for utility of wind power. Wind generating schedules in a wind farm could be efficiently accommodated by means of precise prediction of wind speed to mitigate the impact from instable wind power on power grids. Back-propagation neural network (BPNN) is a main approach for short-term of wind speed prediction. This paper proposes using momentum item to speed up the convergence rate of BPNN, and taking the advantage of the genetic algorithm (GA) to optimize the structure, weights and bias of BPNN. The results of wind speed prediction with real measured wind series by using these three NNs show that GA-BP is a more effective and accurate method to predict wind speed. Finally, the prediction errors in relation with the time scales are discussed.