Abstract: There are many methods of wind power forecasting at present,but each forecasting model has its own advantages and disadvantages,and the reliability of the forecasting results is not high,so it is difficult to meet the requirements of the power grid for wind power forecasting accuracy.In order to deal with this problem,a forecasting model based on the combination of back propagation（BP）neural network and long and short term memory（LSTM） network is proposed for multi-step wind power forecasting,and the feasibility of the combined model in wind power forecasting is verified by simulation.The simulation results show that,the root-mean-square error and absolute average error of the combined model are much smaller than those of the single model of BP neural network and LSTM network,indicating that the combined wind power model is more accurate than the single wind power forecasting model.Therefore,it is suitable for practical multi-step wind power forecasting.