A Method for Predicting Carbon Emission Intensity in Integrated energy system Based on Federated Learning

High precision and granularity carbon emission prediction is crucial for formulating carbon reduction strategies and participating in carbon trading. This article proposes a low-carbon integrated energy system carbon emission prediction method based on federated deep learning. Firstly, to address the issue of insufficient real-time carbon emission data, the production and energy consumption data of corresponding entities in low-carbon operation of the integrated energy system are simulated. Then, in response to the problem of “data silos” among entities in the integrated energy system, a vertical federated learning architecture is established, and short-term carbon emission data predictions are made for each theme based on the LSTM algorithm. Finally, through simulation analysis, it can be seen from the experimental results that the method proposed in this article has the characteristics of high accuracy and strong applicability, which effectively fills the technical gap of insufficient carbon emission data prediction sample data and low prediction accuracy.