Abstract: Establishing the wind power prediction system and improving the prediction accuracy is one of the key techniques for exploiting wind power.Based on numerical weather prediction,a wind power prediction model using the back propagation（BP） neural network is proposed.Factors that affect the prediction accuracy are analyzed using actual data of a certain wind farm.In the light of inconspicuous day characteristic of the original wind speed and the failure of the BP neural network to completely map its power sequence,a prediction model based on wavelet-BP neural network is proposed.With the wavelet-BP neural network model,the wind speed and power sequence are decomposed into different scales.Then the sub-sequences of different frequency components are predicted using multiple BP neural networks.Finally,the output data of BP neural networks are reconstructed to obtain the complete wind power predicting results.It is shown by the research results that the prediction accuracy of wavelet-BP neural network is effectively improved.