Which fossil energy source has the highest average contribution rate to US carbon emissions based on a machine learning algorithm?
Weibiao Qiao, Qianli Ma, Luyao Shi, Haihong Xi, Xinjun Yang, Nan Huang, Yuqin Wang
Abstract
Coal, oil, and natural gas are the main three fossil energy that produce carbon. Among them, which is the main contributor to carbon emissions is rarely studied. In this work, the average contribution rate of carbon emissions is predicted based on an innovative two-stage model combining the optimal layers of wavelet’s orders with long short-term memory optimized by an improved sparrow search algorithm. The experimental results demonstrate that using wavelet for preprocessing can achieve better prediction results, compared to some other preprocessing methods, and the prediction results of one-step prediction are better than those of multi-step prediction. In addition, the six prediction error indicators used in this study are reasonable, and using the average prediction error evaluation indicator is more reasonable. The conclusion can be reached that the order of average contribution rate of carbon emissions from high to low is natural gas, petroleum, and coal and their proportion is 46.62%, 34.90%, and 18.48%, therefore, in the short future, natural gas will be the main source of carbon emissions in the US.