Research on Wind Energy Prediction Based on Deep Learning

The research direction of this paper is to predict wind speed time series based on improved deep learning algorithms. The method in this paper uses Singular Spectrum Analysis (SSA) to decompose the time series, and four methods are used to predict each subsequence. Finally, LightGBM is used to integrate and predict the original sequence. The innovation of this method is to use LightGBM for feature selection and use the selected features for prediction in the final LightGBM model. Compared with the traditional method of using a single model to predict subsequences and then summing the prediction results, the method in this paper can more fully mine the characteristics of wind speed sequences and automatically find a better method than summation for prediction, so as to improve the performance of the model. We tested the method using wind speed data from a wind farm in Weihai City, Shandong Province. The prediction accuracy of the method is 95.5%. The comparison of prediction performance indicators verifies the effectiveness of the method. The method in this paper provides a new idea for the research of Wind Speed Forecasting (WSF) prediction models and has certain guiding significance.