A counterfactual method to measure the potential carbon emission reduction cost of Chinese grid

This study utilizes support vector regression (SVR) technology to establish a relationship model between power cost and various related indicators. The model construction includes data collection, preprocessing, and parameter optimization using Bayesian optimization. The data influencing power costs at different times are normalized and integrated into the model to calculate predicted power cost values before the carbon market commences. The model’s predictions are validated against historical data from 2012 to 2021. By comparing actual and predicted costs before and after the carbon market starts, this study quantifies the impact of carbon market policy on power cost, providing a robust foundation for optimizing carbon emission reduction strategies. This research not only enhances understanding of the correlation between carbon markets and grid carbon emission reduction costs, but also furnishes scientific decision-making support for policymakers.