Analyzing Traffic Conflict Risk at Diverge Area of Left Off-Ramp of Underground Interchanges Using Multinomial Logistic Regression

In order to deeply analyze the traffic conflict risk and the significant influencing factors of conflict risk at the diverging area of the left off-ramp of underground interchanges, this study classifies the conflict risk into three levels: no-conflict risk, general conflict risk, and severe conflict risk, and uses SPSS software to design orthogonal experiments on independent variables and obtain 16 groups of practical solutions, and simulates the experiments by VISSIM microsimulation software, and finally obtains 16916 groups of experimental data and constructs a multinomial disorderly Logit traffic conflict risk model from four aspects, such as the length of the deceleration lane, the traffic volumes of the mainline, the traffic volumes of the ramp and the design speed of the mainline. The results show that, 1) with no-conflict risk as the base group, the four categorical independent variables have a significant effect on both general conflict risk and severe conflict risk when vehicles are driving within the left exit diversion area; 2) when the length of the deceleration lane increases from 120m to 145m and other variables are fixed, the logarithm of the occurrence ratio of the conflict risk of vehicles in the general conflict risk and the no-conflict risk decreases by 0.112; 3) when the traffic volumes of the mainline increases from 5000pcu/h to 5900pcu/h and other variables are fixed, the logarithm of the occurrence ratio of the conflict risk of vehicles in the severe conflict risk and the no-conflict risk increases by 1.192; 4) when the traffic volumes of the ramp increases from 600pcu/h to 650pcu/h and other variables are fixed, the logarithm of the occurrence ratio of the conflict risk of vehicles in the severe conflict risk and the no-conflict risk increase by 1.192; 5) when the design speed of the mainline increases from 50km/h to 70km/h and other variables are fixed, the logarithm of the occurrence ratio of the conflict risk of vehicles in the severe conflict risk and the no-conflict risk increases by 1.679.