Prediction of Carbon Dioxide Emissions in Thermal Power Plants Based on Deep Learning

With the growing global attention to climate change issues, the prediction of carbon dioxide emissions from thermal power plants is crucial for achieving the "dual carbon" goals. This paper proposes a method for predicting carbon dioxide emissions from thermal power plants based on deep learning. A hybrid prediction model is constructed by combining the Logarithmic Mean Divisia Index factor decomposition model, the extended STIRPAT model with deep learning technologies. Firstly, the LMDI model and the extended STIRPAT model are used to analyze driving factors and screen out key influencing factors. Secondly, the collected historical operation data of thermal power plants are subjected to cleaning and standardization preprocessing. Finally, various deep learning models such as LSTM, Transformer and BayesLSTM are constructed and integrated, and combined with uncertainty quantification methods to improve the accuracy and reliability of prediction. This study provides a scientific basis for carbon emission management and policy formulation in thermal power plants.