Study On The Short-Term Wind Power Prediction Method Of The Wind Farm

With the rapid development of wind power generation, installed capacity is increased. Because wind energy is a kind of new energy which is unstable and has stronger intermittent, large-scale wind power grid generation will bring serious impact to power system, mainly in power generation and scheduling plan and system stability, power quality etc. It has the urgent need for wind power to carry on the reasonable forecast in order to reduce the effects of wind power access to power system.This paper briefly introduces the research status of wind power prediction at home and abroad, the application of wind power prediction system for wind power prediction of time series method and RBF neural network basic principle and implementation steps are summarized. Wind power prediction is proposed based on support vector machine (SVM) modeling in the paper. For the parameters in traditional SVM model of wind power prediction are difficult to select problems, this paper proposes a new forecasting model:using improved particle swarm optimization (MPSO) algorithm to seek the optimal parameters of SVM. Wavelet transform (WT) was proposed for decomposition and reconstruction carried out on the wind speed first, then forecast by MPSO-SVM model. At last, the future4hours of wind speed and power are predicted based on WT+MPSO+SVM prediction model, combined with the history of a wind farm in Jiang Xi province wind speed and wind electric power data. Simulation results show that WT+MPSO+SVM model applied to the short-term wind power prediction is effective. Comparing with the methods of time series and RBF neural network, the prediction accuracy improved. Finally visual interface is built for a wind power prediction based MATLAB GUI for more practical of the wind power prediction system.