Mathematical morphology-based short-term wind speed prediction using support vector regression

With the development of the wind power, the prediction of wind speed has received much attention. In this paper, the support vector regression based local predictor (SVRLP) is combined with mathematical morphology, named as SVRLP-MM, and applied to the short-term wind speed prediction. Through the mathematical morphology, the wind speed time series would be decomposed into two subsequences, named “baseline” and “noise”, with different frequencies and wave characteristics. Then the SVRLP is applied to predict each subsequences separately, the final wind speed prediction result is equal to the sum of each subsequences´ prediction result. The proposed SVRLP-MM is evaluated with the real world wind speed data, and is compared with the support vector regression (SVR), the support vector regression based local predictor (SVRLP) and the autoregressive moving average (ARMA). The results demonstrate that the proposed method can achieve a better performance than the other methods.