Abstract: Aiming at the problems of low accuracy and poor generalization of wind speed forecast due to randomness and volatility, a combined prediction model based on variational mode decomposition（VMD）, whale optimization algorithm（WOA）, long short-term memory neural network（LSTM） and sparrow search algorithm（SSA） was proposed. Firstly, WOA is used to automatically optimize the core parameters of VMD（K value and penalty coefficient α）. After decomposing the wind speed time series, SSA is introduced to optimize the core learning parameters of LSTM, and finally, the predicted wind speed data of each subcomponent is integrated to obtain the final predicted wind speed, which is verified by a number of model evaluation indicators. The RMSE, MAE, MAPE and R2 of the model are 0.0758 m/s, 0.0578 m/s, 1.492% and 0.979, respectively. Compared with other single optimization prediction models WOAVMD-LSTM and VMD-SSA-LSTM, the relevant evaluation indicators have significantly improved.