A Carbon Emission Projection and Prediction Method Based on LSTM and ARDL Models for Multiple Emission Sources

To enhance carbon monitoring and calculation systems across various temporal and spatial scales, it is imperative to strengthen the digital infrastructure for electricity-to-carbon conversion and promote the quantitative integration of physical and digital carbon reduction efforts. This necessitates a concentrated focus on key technologies related to carbon metering, the conversion of electrical energy into carbon emissions, and comprehensive carbon emission accounting across different entities. This paper leverages the high frequency, accessibility, real-time capabilities, and accuracy of extensive power sector data to investigate the coupling relationships among electricity, energy, and carbon. Utilizing advanced big data analytics, we develop a model of the electricity-energy-carbon relationship to address the research needs of carbon emission entities. This model facilitates the calculation, monitoring, analysis, and prediction of carbon emissions, even in scenarios with limited measurement data.