Abstract: At present, the global demand for renewable energy is growing, and wind energy as a renewable clean energy plays an important role in the power supply.However, the fluctuation of wind power brings challenges to power grid dispatching.Accurate prediction of wind power becomes the key to optimize power system operation.Traditional forecasting methods, such as physical modeling and statistical modeling, have problems of complex calculation and high data quality requirements.Based on this, a wind power prediction method based on random forest (RF), particle swarm optimization (PSO) algorithm and long short-term memory (LSTM) neural network is proposed.Firstly, the wind power curve is used to preprocess the data to eliminate the outliers.Secondly, RF algorithm is used for feature selection to remove unimportant feature variables, and PSO algorithm is used to optimize LSTM hyperparameters to improve the prediction accuracy of the model.Finally, the open data set of a wind farm in the southeast coastal area of China is used to conduct model experiments and ablation experiments.The experimental results show that the root mean square error (RMSE), mean absolute error (MAE) and mean absolute percentage error (MAPE) of the test set are 1.512 8, 1.163 2 and 8.982%, which are significantly better than the single LSTM model and other combined models.It provides an efficient and accurate method for wind power prediction, and provides a strong support for wind farm optimization scheduling.