Short-term wind speed prediction of wind farms based on improved particle swarm optimization algorithm and neural network

摘要

Short-term prediction of wind speed is important to the power system operation. Generation schedules in a wind farm could be efficiently assigned by means of precise prediction of wind speed to alleviate the impact of instable wind power on power grids. Back-propagation neural network (BPNN) is a main approach for short-term wind speed prediction. The method, using the improved particle swarm optimization (IPSO) algorithm to train the BPNN, is proposed in this paper. The model of wind speed prediction also takes meteorological factors like temperature into consideration. The performance of BPNN, PSO based BPNN and IPSO based BPNN for one-hour ahead forcasting of wind speed have been examined with real data. Simulation results clearly indicate the advantage of IPSO based BPNN over the other two methods in convergence speed and prediction precision.