Research On The Influencing Factors And Forecast Of Carbon Emissions In Tianjin’s Transportation Industry

Promoting the construction of a strong transportation country and building a modern and high-quality national comprehensive three-dimensional transportation network are important objectives for the development of China’s transportation industry in the current and future periods.Analysing the current situation of China’s transport industry and accurately predicting its carbon emissions will be conducive to enhancing the green development of national transport and building a green and low-carbon transport development model.Based on the current situation of energy consumption and carbon emission of transportation industry in Tianjin,this paper analyses the factors influencing carbon emissions in the transport sector from various perspectives,such as macroeconomic factors,industry development factors and energy-related factors.The paper also uses the information gain method to rank and filter the factors influencing carbon emissions in Tianjin’s transport sector,and finally concludes that four indicators-transport infrastructure investment,resident population,transport energy intensity and energy supply-are the key factors influencing carbon emissions.They were used as input indicators for the carbon emission portfolio weight prediction model.Secondly,based on the current situation that deep learning is rarely used for carbon emission prediction in the transportation industry,three individual prediction models,LSTM,GRU and Bi-LSTM,are constructed based on deep learning neural networks,and each individual model is trained and tested to obtain the optimal architecture.A comparison of the individual models shows that the Bi-LSTM has the lowest prediction error and is suitable for predicting carbon emissions in the transport sector.Two combined weight prediction models were also constructed based on the single best model and different weight determination methods,and the weight models can well combine the advantages of each individual model.The experimental findings show that the prediction error of the weight model is substantially improved,and the prediction accuracy is 2.89% and 2.67% higher than that of the Bi-LSTM model,respectively.Finally,based on the prediction results of the combined weight prediction model and the current situation of the actual development of the transport industry in Tianjin,energy saving and emission reduction strategies are proposed in terms of population quality,consumption structure,transport structure,transport efficiency,energy consumption structure and energy use efficiency.