Study On Wind Speed Forecasting Method For Wind Farm

Global non renewable energy shortage, wind energy, as an environmentally friendly energy, has become an important alternative to fossil fuels.But the wind has the characteristics of intermittent, volatility and randomness. Wind power has brought some challenges to the stability of power system. The prediction of wind power is an important way to solve this problem. Wind speed forecasting is an important part of wind power prediction. Under this background, this paper focuses on the following aspects of the short-term wind speed prediction:BP neural network, wavelet neural network and support vector machine(SVM)prediction model are established, and the wind speed of wind field is predicted, then the error analysis of the prediction results of each method is carried out. The analysis of measured wind speed data shows that SVM has higher prediction accuracy compared with the other two methods. Therefore, this paper will SVM as the basis of the combination model.According to the non-stationary characteristics of wind speed time series, a combined model is proposed to predict the wind speed. The method first uses EEMD will wind speed time series is decomposed as a group with the components of different frequency components; then will be close to the sample entropy of each component superimposed.Finally, SVM to reconstructed sequences were predicted and the prediction results are superimposed. The analysis of measured data shows that the combined model has better prediction accuracy than the SVM method and EEMD-SVM method..A rolling multi-step prediction model based on combination model is established.Firstly, based on the direct multi-step prediction model of SVM and the rolling multi-step prediction model based on SVM, the wind speed of the wind field is predicted. The error analysis shows that the precision of the rolling multi-step prediction is higher than that of the direct multi-step prediction. The error analysis shows that the combination model has higher prediction accuracy with the rolling prediction method based on SVM and the direct multi-step method based on SVM. It proves that the method is feasible for the multi step prediction of wind speed.