An Optimization Framework for Monitoring and Predicting Electricity Carbon Emissions Based on Sparrow Search Algorithm and Improved BPNN Neural Network

In recent years, abnormal weather has occurred frequently, and effectively controlling and reducing carbon emissions has become one of the important challenges currently faced by society.With the acceleration of electricity grid construction, a large number of new energy projects are gradually connected to the grid, and the electricity grid system is developing toward a clean and low-carbon direction.The prediction of power carbon emissions has enabled quantitative research on power system carbon emissions, and understanding the changing trend of power carbon emissions is of great significance for promoting the decarbonization of power systems.This study constructs a PE-PACF-SSA-BPNN combined prediction model for electricity carbon emissions.First, external factors affecting electricity carbon emissions are selected based on the Pearson coefficient (PE), and partial autocorrelation analysis (PACF) is conducted on the electricity carbon emission sequence.The internal influencing factors of electricity carbon emissions are selected based on partial autocorrelation coefficients, which reflect the changing trends and patterns of electricity carbon emissions.Second, a BPNN neural network is used to model the correlation between electricity carbon emissions and influencing factors, and the BPNN parameters are updated through an error-propagation mechanism.Carbon emission prediction is realized based on influencing factors and the BPNN neural network.Simultaneously, the Sparrow Search Algorithm (SSA) is used to optimize the weights between the input layer and the hidden layer of the BPNN to improve model robustness.The prediction errors of this model in four scenarios -China, Guangdong, Shandong, and Jiangsu -are 0.628%, 2.924%, 1.852%, and 1.321%, respectively.This research constructs a novel tool and method for carbon emission Pol.J. Environ.Stud.Vol.XX, No. X (XXXX), 1-16 prediction in the power industry, which is helpful for the tracking and prediction of the carbon footprint of the power industry and provides a reference for its clean development.