Developing carbon mitigation measures requires information.In order to obtain necessary auxiliary information,the dynamic time warping cluster method is first used to divide 30 provinces in China into four types based on the similarity of CO2 emissions;Then for each emission category,the critical driving emission sources of the total emissions are identified us-ing LASSO regression;Finally,the emissions in four scenarios are predicted using long short-term memory and gated recurrent unit neural network models,and the"carbon peak"path is explored.The results showes that the critical emission sources of Types 1 and 2 are diesel con-sumption in the transportation industry and coal consumption in the power&heat industry;the critical emission sources of Type 3 are coal consumption in the power&heat Industry and coke consumption in the steel industry;The key emission sources for Type 4 are gasoline in the trans-portation industry,raw coal in the power generation and heating industry,and the production process of non-metallic mineral products in the sector.In terms of peak tasks,type 3 provinces have more challenging tasks than other types of provinces.
关键词
省级CO2排放/聚类分析/神经网络模型
Key words
Provincial co2 emission/Clustering analysis/Neural network model
万方数据