To quantitatively measure the carbon emissions of road transportation in mountainous cities and its influencing factors,this paper uses portable emission measurement system to collect the carbon emissions of cars in different time and space,and selects eight variables in three dimensions,including vehicle engine characteristics,driving characteristics and road traffic environment,which are explanatory variables in combination with the road traffic characteristics of mountain cities.Utilizing the random forest regression model,this paper developed an analytical model to understand the factors influencing carbon emissions from road transportation in mountainous cities.The partial dependence function was used to quantitatively analyze the non-linear and interactive effects of these factors on carbon emissions.Additionally,a comparative analysis was conducted across multiple models.The results indicate:(1)The random forest regression model has better performance and prediction accuracy than the multiple linear regression,support vector machine and Long Short-Term Memory Network(LSTM)models,with a goodness-of-fit of 74.3%.(2)From the perspective of impact contribution,speed(25.97%)and exhaust cylinder temperature(23.73%)are the most important factors affecting carbon emission factors.(3)The nonlinear effect of velocity and elevation change on carbon emission factor is significant,and the threshold of speed is 58 km·h-1 and the threshold of elevation change is 0.10 m.Complex traffic structures such as overpasses and steep slopes in mountainous cities should be optimized to reduce the range of vehicle speed fluctuations.(4)There is a positive interaction between low-speed driving and other factors.The research results can be combined with real-time calculation of road traffic carbon emissions based on road traffic volume and road condition information,providing decision-making basis for traffic management strategy design.
维普
万方数据