Research on wind power load forecasting based on deep learning

Le résumé fourni par la source

To accurately predict wind power load, we need to understand the time series and environmental conditions. A new forecasting model based on Bidirectional Long Short-Term Memory (BiLSTM) can help. The correlation of the wind power load data set is analysed by Spearman's correlation coefficient, the data is extracted by multi-scale temporal sequence, and the temporal features are filtered and screened by the attention mechanism, and finally the predicted values are integrated and output by the fully connected layer. Algorithms and parameters are being optimized, including Particle Swarm Optimization (PSO), comparative experiments are carried out. In this paper, the 1-year multidimensional wind power load data of a region is used as the data set for training and validation, and the advantages of the PSO optimisation algorithm combined with this composite model over the other algorithms and parameter conditions in terms of prediction accuracy and stability are finally verified.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.