Abstract: Aiming at the strong nonlinearity of wind speed,a short-term wind speed prediction model based on singular spectrum analysis and improved particle swarm optimization adaptive fuzzy inference system is proposed. The method uses singular spectrum analysis to decompose the original series into trend and harmonic components,then establishes a fuzzy neural network model for each component,and finally superimposes each component prediction result to obtain the next wind speed value. In order to improve the prediction accuracy,the improved particle swarm optimization algorithm is used to optimize the membership function of the adaptive fuzzy inference system. Simulation based on measured data from a wind farm in Hebei Province and comparison with traditional neural networks. The results show that the re-superimposition of the wind speed after re-construction reduces the complexity of the original problem and improves the prediction accuracy. This model significantly reduces errors in multi-step predictions in wind speed sequence predictions at different time intervals.