Wind power prediction method based on CNN-LSTM-Attention and Bayesian optimization

Wind power generation prediction plays an important role in the operation of power system. Due to the complexity and randomness of wind speed, the prediction accuracy of current short-term wind power prediction models is reduced. This paper proposes a combination model based on convolutional neural network (CNN), long short-term memory network (LSTM) and attention mechanism. Convolutional networks are mainly used to extract spatial features of time series data, and long short-term memory networks are mainly used to process time series modeling of time series data. The attention mechanism is used to focus weighted sequences output by the LSTM layer to extract the most relevant and important time step information. In order to improve the accuracy of wind power prediction results, the combination model is optimized by Bayesian Optimization algorithm to find the best hyperparameter combination. The test results show that the proposed method can significantly improve the accuracy of wind power prediction.