Study on Wind Power Prediction Based on EEMD-LSTM

Accurate prediction of wind power generation is important to achieve user-friendly interconnection of large-scale wind power generation networks. Given the intermittency and instability of wind power, an ensemble empirical mode decomposition (EEMD) and long and short memory neural network (LSTM) method is proposed to predict wind power generation. First, the original wind power sequence was decomposed into multiple intrinsic mode functions (IMFs) using the EEMD method, then LSTM-based prediction models were built for each IMF separately to obtain prediction results for each subset component, and finally the prediction results for each subset were superimposed to reconstruct the resulting wind power sequence. The case simulation results showed that the EEMD-LSTM prediction model had higher prediction accuracy than the LSTM prediction model and the EMD-LSTM prediction model, and three evaluation indicators, RMSE, MAE and MAPE, were superior, validating the accuracy and superiority of the combined EEMD-LSTM prediction method.