Deep Learning-Based Comparative Modeling of Carbon Emissions Projections

Abstract

Scientific statistics of carbon emission data and reasonable prediction of its development trend can help stabilize carbon emission to ensure that China can achieve the carbon peak by 2030. In this paper, we use Pearson correlation analysis to select the main factors that optimally affect carbon emissions and predict them through the back propagation neural network model optimized by whale algorithm, the BP neural network model optimized by genetic algorithm and the back propagation neural network model, so as to provide an effective prediction method for scientists to study the growth rate of carbon emissions production in China The back propagation neural network model is used to predict the carbon peak and the time of carbon neutralization in China by the optimal model. The Pearson correlation coefficient screening was performed with 85 groups of factors affecting carbon emissions collected in China from 1998–2019 under carbon emissions and (energy, agriculture, industry, integrated, etc.) resource environment, and eight optimal groups of data were used for prediction. The data were compared between R Square, MSE, RMSE and MAPE data by back propagation neural network, optimization using whale algorithm and optimization using genetic algorithm, and the optimal model was used to predict carbon emissions for three years from 2020 to 2069. The results of this study show that the four groups of data, R Square, MSE, RMSE and MAPE, predicted by the BP neural network model optimized based on the whale algorithm are the best among the three models, and the data from 2016 to 2069 can be accurately predicted by the whale algorithm optimized BP network to determine the time of reaching carbon peak and carbon neutrality in China. Compared with other prediction models, the BP neural network model optimized by the whale algorithm can effectively predict carbon emissions and provide an optimal method for carbon emissions prediction.