摘要
Accurate prediction of wind power is significant for power system dispatching as well as safe and stable operation. By means of BP neural network, radial basis function neural network and support vector machine, a new combined method of wind power prediction based on cooperative game theory is proposed. In the method, every single forecasting model is regarded as a member of the cooperative games, and the sum of square error of combination forecasting is taken as the result of cooperation. The result is divided among the members according to Shapley values, and then weights of combination forecasting can be obtained. Application results in an actual wind farm show that the proposed method can effectively improve prediction precision.
Accurate prediction of wind power is significant for power system dispatching as well as safe and stable operation. By means of BP neural network, radial basis function neural network and support vector ma- chine, a new combined method of wind power prediction based on cooperative game theory is proposed. In the method, every single forecasting model is regarded as a member of the cooperative games, and the sum of square error of combination forecasting is taken as the result of cooperation. The result is divided among the members according to Shapley values, and then weights of combination forecasting can be obtained. Application results in an actual wind farm show that the proposed method can effectively improve prediction precision.
