Regional Carbon Emission Prediction and Low-carbon Path Analysis Based on BP Neural Network Model

Abstract

To attain carbon emission control and sustainable economic development, tailored low-carbon policies must be adapted to distinct regional contexts. Given disparities in key industries, economic growth, and resource availability, variations in carbon emissions across China’s eastern, central, and western regions necessitate divergent low-carbon strategies. To comprehensively grasp regional carbon emissions and provide precise recommendations, a Lasso regression method identified five influential factors from seven, including population, per capita GDP, total energy consumption, energy mix, industrial makeup, urbanization rate, and forest coverage. Analyzing the link between carbon emissions and total output, a predictive model employed a GAoptimized BP neural network to forecast emissions. Findings indicate higher carbon emissions in the developed eastern region due to rapid economic growth, industrial production, and energy use. While the east region maintains emission leadership, emission growth rates converge, reflecting nationwide progress in reduction efforts. Future strategies should focus on regional development, exploring low-carbon paths through energy restructuring, urban optimization, and synergy of energy strengths, thereby achieving shared lowcarbon objectives.