A nonlinear combined model for wind power forecasting based on multi-attribute decision-making and support vector machine

Given the significant fluctuation of errors for single forecasting model and limitation of linear combined forecasting models, a nonlinear combined model for wind power forecasting based on multi-attribute decision-making and support vector machine (SVM) is proposed. Firstly, based on the theory of multi-attribute decision-making, the best three models are chosen as single forecasting models to verify the forecasting results. Three different results are obtained with each modeling and forecasting method separately. Then, SVM combined forecasting model is built by use of all single models as training inputs and corresponding actual values as training outputs. In order to validate the proposed model, two different historical data are used. The results show that the combined model effectively improves forecasting accuracy with synthesizing the advantages of all single forecasting models. Both root-mean-square error and mean percentage error are better than single models and other combined models. Finally, the effect of sampling interval on the forecasting results is studied, and the result indicates that the forecasting accuracy is more superior when sampling interval is between 5 minutes to 15 minutes. © 2013 State Grid Electric Power Research Institute Press.