In order to enable the wind power system run smoothly and reduce the economic loss caused by power system fluctuation, and at the same time improve the competitiveness of the wind power system, it is of great and practical significance to find a stable and accurate wind speed prediction method. Among the machine learning methods, a neural network based on the back propagation (BP) algorithm to adjust the weight is one of the most commonly used and most effective methods. Although the BP neural network has a strong ability to fit nonlinearity, it converges slowly in the process of adjusting weight and easily falls into local optimal values. In order to effectively solve these two problems that will occur in the prediction process of the BP neural network, genetic algorithm (GA) was used to optimize the neural network in this paper. On this basis, considering the intermittency, non?stationarity and difference of wind speed series, a short?term wind speed prediction model EMD?GA?BPNN based on empirical mode decomposition (EMD), genetic algorithm (GA) and BP neural network was proposed. The reliability and advantages of this model in short?term wind speed prediction were verified by transverse comparisons with other models.