Short-term Wind Power Prediction by Hybrid Neural Network Based on Information Decomposition

In response to China’s goal of achieving carbon neutrality in 2060, the Chinese government vigorously promotes the spot energy market, and the power generation prediction of new energy participation in the spot energy market is an urgent problem to be solved. In this paper, the Numerical Weather Prediction (NWP) data and the historical wind power data are used to predict the day ahead of wind power. The strong randomness of wind power leads to low accuracy of ordinary prediction methods. In this paper, the historical wind power data was decomposed by using the Complementary Ensemble Empirical Mode Decomposition (CEEMD) signal decomposition method. Decompose into high and low frequency signals, and add NWP weather data after decomposition. Convolutional Neural Networks (CNN) are used to extract data spatial features and input them into Long Short-Term Memory (LSTM) Neural network. The accuracy of wind power day-ahead prediction is improved by time feature extraction and effective gating structure. Through experimental verification and comparison, the Mean Absolute Percentage Error (MAPE) and Root Mean Square Error (RMSE) of the proposed method are lower than those of other methods. It is proved that the method presented in this paper can effectively improve the accuracy of day-ahead wind power prediction.