Research On Influence Factors And Prediction Models Of Carbon Emissions From Provincial Transport Industry

China has set a goal to achieve carbon peaking by 2030.As one of the fastest-growing sectors of China’s economy,the transportation sector also accounts for a particularly significant proportion of carbon emissions in emissions from energy activities,and the share of carbon emissions from China’s transportation sector in global transportation carbon emissions also reached 11%in 2019,ranking second in the world.Therefore,this paper takes Hainan Province as an example to accurately identify the influencing factors of the provincial transportation sector and clearly forecast the carbon emissions of the transportation sector,so as to lay a theoretical foundation for promoting the green and low-carbon development of the transportation industry and accelerating the service of the carbon peak target.Firstly,this paper analyzes the current situation of transportation industry and energy consumption in Hainan Province,and uses Tapio decoupling model to quantitatively analyze the measured carbon emissions of transportation industry in Hainan Province.The study shows that the carbon emission of transportation industry in Hainan Province shows an overall increasing trend but the growth rate tends to be close to 0.The growth rate of carbon emission of transportation industry in Hainan Province is getting slower and slower relative to the economic development rate,and most of them show weak decoupling in recent years.Secondly,combining with related literature and constructing the extended STIRPAT model,a total of 8 kinds of 12 influencing factors were selected from the four perspectives of population size,economic development,technology and transportation.At the same time,RF-VIM was used to filter the influencing factors and combined with the results of Pearson correlation analysis to select 9 influencing factors,which are energy consumption in transportation industry,GDP per capita of Hainan Province,year-end population of Hainan Province,gasoline energy consumption ratio,cargo turnover,passenger turnover,natural gas energy consumption ratio,diesel energy consumption ratio and kerosene energy consumption ratio.And they are entered as variables in the prediction model later.Finally,based on the previous deep learning methods in carbon emission prediction,three improved LSTM models,CNN-LSTM,CNN-Bi LSTM and CNN-Bi LSTM-Attention,were constructed and compared with LSTM models.The CNN-Bi LSTM-Attention model has the highest prediction accuracyR ~2of 0.926.The CNN-Bi LSTM-Attention model is selected to predict the transportation carbon emissions in Hainan Province,and the expected carbon peak time is 2028.