Abstract. This paper aims to establish a suitable wind power forecasting model used to forecast the power of the wind farm. An improved real-time series model is built by linear function and Fourier function. For raw data, the picture of historical data is adopted to correct them. In order to improve the prediction accuracy of wind power, it proposes the linear combination model based on improved time series model and grey model. The model uses a fixed weight method. Mathematical analysis and calculation results show that combination model, which has certain reference value, is simple and can improve the prediction accuracy overall.