货车移动遮断影响下的跟驰风险异质性建模

Modeling Heterogeneity for Car-following Risk Evaluation Under Truck Movement Block

摘要:

基于移动瓶颈理论和交通流理论构建“货车移动遮断”效应模型，解析货车移动遮断形成机理，选用无人机采集货车移动遮断场景中车辆行驶视频数据并提取高精度车辆跟驰轨迹样本，基于此提出考虑冲突可能性和冲突严重度的小客车跟驰风险评价方法和分级标准，利用RP-ORP模型构建了考虑异质性跟驰风险概率预测模型。结果表明：货车移动遮断动态影响交通流稳定性，其形成过程包括减速跟驰和加速超车两个阶段；考虑异质性的RP-ORP模型能实现特定条件下小客车跟驰行为处于不同风险等级的概率预测，且拟合优度高于FP-ORP模型高；货车纵向加速度、跟驰车头间距、跟驰持续时间、小客车与货车速度差、激进型驾驶员5个变量显著影响小客车跟驰风险水平，且跟驰持续时间和激进型驾驶员2个变量具有随机参数特性。

Abstract:

Based on the movement bottleneck and traffic flow theory， a "truck movement block" effect model was constructed to analyze occurrence mechanism and the car-following behavior. Then， the UAV was selected to collect the vehicle driving video under the truck movement block， and the high-precision car-following trajectory was extracted by trajectory software. Last， the car-following risk assessment methods and grading standards that considers both the possibility and the severity of the risk was proposed. And the random parameter ordered probit model （RP-ORP） was established to assess the car-following risk， and the elasticity coefficient was introduced to analyze the magnitude and direction of the impact on car-following risk. This finding demonstrates that truck movement block includes two processes： car-following and overtaking， which dynamically affects the traffic stability. the RP-ORP model has a better fit than the FP-ORP model， and can realize the probability prediction that different car-following levels under certain conditions. Five variables significantly affect the car-following risk level， such as truck longitudinal acceleration， car-following distance， car-following time， average car-following speed difference， radical driver， among the variables of car-following time and radical driver have random parameter characteristics.