Research on short-term wind speed prediction based on EEMD-GRU prediction model

Accurate wind speed and power prediction can lead to a stable large-scale wind power grid connection and improved wind energy utilization. As a result, this paper proposes a model for predicting short-term wind speed based on the Ensemble Empirical Decomposition Model (EEMD) of white noise sequences with normal distribution and the Gated Recurrent Unit (GRU) of recurrent neural networks. First, the original sample data of wind speed are divided into several subsequences using EEMD. The deconstructed sub-sequences are then each predicted using GRU to produce numerous sub-prediction results. Finally, the sub-prediction results are superimposed to get the final prediction results, and the accurate prediction of wind speed can be realized. The experimental results indicate that the MAE, RMSE, MAPE and R
²
of the proposed model are 0.11998, 0.16233, 1.77205 and 0.98827, respectively. So the conclusion that can be drawn is that this model has better prediction results than the comparison model.