Multi-source data-driven technology research on carbon emission dynamics prediction in electric power industry

Abstract

Abstract This paper analyzes the trend of power generation structure and carbon emission changes in the power industry and decomposes and analyzes the influencing factors of carbon emission in the power industry by using the LMDI decomposition method. Combined with the analysis of the influencing factors of carbon emissions in the power industry from 2016 to 2022, the carbon emissions of the power industry in the Yellow River Basin are simulated by the scenario analysis method. Four simulation scenarios were constructed based on the economic scale, industrial structure, industrial electricity consumption intensity, thermal power fuel conversion rate, and power supply structure. The IPSO-LSTM model for carbon emission prediction was created after optimizing the LSTM neural network prediction model. Combining the scenario analysis method to set the amount of changes in the high carbon, baseline, and low carbon scenarios of the influencing factors, the carbon emissions from the power sector in different scenarios are predicted for the years 2025-2035. From 2025 to 2035, the carbon emissions from the power sector in the three scenarios, except for the energy transition scenario, show a trend of increasing, then decreasing, and then increasing over the study period. The energy transition scenario shows a pattern of increasing and decreasing carbon emissions from the power sector.