Research on the application of deep learning algorithm in energy management for low-carbon society

Abstract The present study endeavors to investigate the application of deep learning algorithms in energy management, aimed at fostering a low-carbon society. Specifically, the study focuses on the performance of wavelet packet decomposition for noise reduction in predicting carbon emissions. The results indicate a significant enhancement in predictive accuracy, outperforming models without noise reduction across various evaluation metrics. This research not only elevates the precision and reliability of predictions but also underscores the critical role of data preprocessing in complex tasks, thereby offering novel methodologies and perspectives for carbon emission monitoring and forecasting in energy management.