Research On Wind Speed Modeling And Prediction Of Wind Plant

In order to eliminate the adverse effects brought by large-scale development of wind power on grid stability, short-term wind speed and wind power forecasting of wind farm has become the focus of attention at home and abroad. It can adjust the scheduling, effectively mitigate the adverse effects of wind power on the grid, reduce the power system operating costs and spinning reserve. A the same time it also can improve the competitiveness of the wind farm in the electricity market.This article describes the knowledge of the wind speed and combine with the whole year’s wind speed data measured by wind tower on a wind farm for wind speed characteristics analysis. Least squares support vector machine (LS-SVM) has predicting and fitting abilities to nonlinear problems. Short-term wind speed forecasting model is established based on historical data. According to the characteristic of wind, the LS-SVM model based on similar data and wavelet analysis is proposed to improve the prediction accuracy of the LS-SVM model. The affects of model prediction accuracy brought by similar data and wavelet analysis are analyzed. Simulation results showed that the induction of similar data and wavelet analysis improved the prediction accuracy. The model has better generalization capability. Not only the run time was not increased, but also project needs were meet. LS-SVM parameter selection methods were analyzed. Particle swarm optimization (PSO) is used to optimize the LS-SVM model regularization parameter C and kernel function parameter r. And comparative analysis is made between PSO and the grid search method.