Applied Research Of Support Vector Machine On Forecasting Of Wind Speed In Wind Farm

In order to eliminate the adverse effects to power grid stability by developmenting the wind power in large-scale, The short-term forecasting of wind speed and the wind turbine generated energy has become a common concern at home and abroad. The accurately predict of wind speed can adjust scheduling plan, can reduce the adverse effects of power grid effectivly,at the same time the competition ability of wind farm has been improved in power markets. Based on the wind speed data in Inner Mongole wind farm character analysis, Then directly forecast the short-term wind speed by the least square support vector machine (SVM) method. And, analysis the parameters of LS-SVM, The genetic algorithm is adopted for the LS-SVM.the regularization parameters and kernel function is optimized, and the optimized LS-SVM model is used for the short-term wind speed forecasting, compare to the gridsearch optimization method, the results showed that the optimization model by genetic algorithm (GA) is superior to the gridsearch optimization method. At last,on account of non stationarity of the wind speed,the wavelet transform is used to pretreatment the wind speed.and WT&LS-SVM combined forecasting model. Test results show that the effect of WT&LS-SVM model is best, Consequently the accurate predicting of the wind speed effectively lighten the adverse effect to the whole power grid by the wind speed fluctuation.