Fusion modeling for wind power forecasting based on redundant method elimination

In general, an individual forecasting model for wind power cannot always attain a sufficient degree of precision. To further improve the forecasting accuracy, this paper develops a new wind power forecasting model using a fusion model based on the concept of optimal selection and the Shapley Value. The fusion modeling method first establishes an individual forecasting model base. Next, higher-precision models are optimally selected using grey incidence analysis based on the collective evaluation index, and the redundancy of these models is judged on the basis of the information matrix of forecasting errors. This allows redundant models to be eliminated so as to simplify the establishment of the fusion model. Eliminating redundant models avoids negative degree of fusion. Finally, the fusion degree of each optimally selected individual model is calculated according to the Shapley Value. Models, which include the proposed model, seven individual models, and two other combined models, are tested to forecast t...