Aiming at the limitations of single forecasting method, the improved combination prediction model based on IOWGA operator for degree of logarithm grey incidence is used to build an optimal combined forecasting model for wind power, and the model is optimized by improved multiple population genetic algorithm (MPGA) RBF neural network method, similar day method and support vector machine (SVM) method are respectively used to predict wind power for predicting daily and the day before, the optimal combined forecasting model is used to predict 24 h wind power for predicting daily. The actual example is analyzed, according to the measured data of a wind farm in Yunnan province, the results show that the optimal combined forecasting model could effectively improve the forecasting accuracy for wind power, and has stronger practicality.