基于CEEMD-CNN-GRU的风电功率预测OA

The traditional prediction model in wind power prediction is difficult to fully extract the spatio-temporal characteristics and hidden features in the wind farm historical data,and the prediction accuracy is low.Aiming at this problem,a CEEMD-CNN-GRU wind power prediction model has been proposed.Firstly,the power sequences are decomposed using complementary ensemble empirical modal decomposition(CEEMD)for different wind power scenarios to reduce the volatility of the wind power sequences.Then,the spatial features are extracted using the convolutional neural network(CNN),and the temporal features are extracted using the gated recurrent unit(GRU).Lastly,the wind power prediction is achieved,and the decomposed sequences are superposed to obtain the final prediction results.The results show that the designed model is highly accurate,and the root mean square error is reduced by 80.17%,77.07%and 71.07%compared to the CNN,GRU and CNN-GRU models,respectively.Moreover,the root mean square error is reduced by 78.63%and 66.61%after turbines are grouped and scenarios are divided compared with those without turbines grouping or scenarios dividing,respectively.