Short-term Wind Power Prediction Method Based on GCN-LSTM

Led by the strategic goal of "carbon peaking and carbon neutral", the development of new energy, especially wind power, will occupy an important position in the power grid. The current wind power prediction has a large error, which has a serious impact on the safe and economic operation of the grid. In this paper, a combined prediction model combining graph convolutional network (GCN) and long short-term memory network (LSTM) is proposed for short-term wind power prediction by combining deep learning. The method firstly uses graph convolutional network to extract spatial relationship features of wind power historical data; secondly, it uses long and short-term memory network model to extract and predict the temporal features of wind power. According to the experimental results, the method has less error and higher accuracy than the single-model method in prediction results.