Probability Density Prediction of Wind Power Based on BiLSTM-DCC

With the large-scale integration of fluctuating and uncertain wind power into the grid, predicting wind power as accurately as possible can help improve the stability of the power system. In this paper, we combine the advantage of bidirectional long and short-term memory (BiLSTM) neural network and dilated causal convolutional (DCC) neural network, and proposed a probability density prediction method for wind power based on BiLSTM-DCC. Firstly, the wind power data are preprocessed. Then, a BiLSTM-DCC model is used for prediction, where BiLSTM is able to mine the temporal features of wind power data. Finally, the Gaussian kernel density estimation algorithm is used to obtain the wind power probability density prediction results. Simulation results show that the proposed method can improve the accuracy of wind power probability density prediction.