Carbon Dioxide Emissions Prediction in China: A Genetic Algorithm-backpropagation Neural Network Approach

Abstract

Abstract With the global climate problem worsening, accurately predicting carbon dioxide emissions has become critical. In this study, we compared and selected various neural network models and a support vector regression (SVR) model to predict China's carbon dioxide emissions from 2022 to 2030. Among the models tested, we found that the Genetic Algorithm-backpropagation (GA-BP) neural network model had the highest prediction accuracy and therefore used it to predict China's future carbon dioxide emissions.We also used the GA-BP neural network model to analyze the factors influencing carbon dioxide emissions and compared the results under three scenarios: baseline, low, and high. Our findings show that under the low carbon emission scenario, China is projected to reach its carbon peak in 2028, with a carbon dioxide emission of 12.184 billion tons. The baseline scenario is projected to reach its carbon peak in 2029, with a carbon dioxide emission of 12.291 billion tons. However, under the high carbon emission scenario, China is not projected to achieve a carbon peak by 2030. Overall, our study provides insights into China's future carbon dioxide emissions, which can inform policy decisions to reduce carbon emissions and mitigate the effects of climate change.