Short-term wind power prediction by using empirical mode decomposition based GA-SYR

Accurate short-term wind power prediction can improve the trade and the dispatch level of wind power. To predict the short-term wind power, we investigated the empirical mode decomposition (EMD) of numerical weather prediction (NWP) and genetic algorithm (GA) optimization of support vector regression (SVR). First, the wind speed data from NWP is decomposed into the EMD components, including multiple intrinsic mode functions (IMFs) and one residue. Then, a genetic algorithm-support vector regression model (GA-SVR) is used to build models of all components. Finally, the prediction results of all components are integrated, which as input of wind-power curve for obtain the predicted value of wind power. In order to validate the proposed method, one real-world dataset was used for model training and testing, and the results show that the proposed method is more accurate and effective than SVR and back propagation neural networks (BPNN).