Wind power forecasting is vital to the stable operation and economic dispatch of power system. In order to fully mine the effective information in historical data to improve the accuracy of short-term wind power prediction, a short -term wind power prediction method based on convolution neural network (CNN) and long short-term memory network model(LSTM) is proposed. The effective information was extracted by CNN sequence feature extraction ability, and the data was input to LSTM network after other information was removed. The problem of gradient dispersion could be solved by keeping longer effective memory information, which made up for the lack of instability and gradient disappearance when the LSTM network model was faced with a long sequence. The results show that the method proposed is more accurate than the back propagation neural network and LSTM network.