Accurate wind power prediction is an important method to guarantee the power system containing large-scale wind power to be safe and stable. Using a single prediction model to predict will all have their own advantages and disadvantages. Therefore, in order to improve the prediction accuracy of wind power prediction, the paper puts forward a combination method for wind power prediction based on improved entropy method, and uses the rolling weight at the same time. In this way, it achieves the complementarity of all single prediction models. Taking the real-measured data from a wind farm in the west of Jilin province as an example, case study is done. It illustrates that improved entropy method is effective when it is used to predict the wind power. In the same time, it is proved that the rolling weights can realize the constant renewal of weights, and reflect the latest changes of wind power, so as to achieve the increase of wind power prediction accuracy.