高速公路常发拥堵路段追尾事故风险实时预测

Real-time prediction of rear-end crashes near freeway recurrent bottlenecks

研究了个体车辆遇到运动波(Kinematic waves)前后行驶轨迹的特征,分析了追尾事故发生条件,将多个车辆轨迹进行集计分析,提出了基于集计交通流数据的追尾事故风险预测模型。结果表明:运动波传播过程中追尾事故概率与事故风险指数及上游占有率标准差显著相关;采用模型对实际高速公路常发拥堵路段追尾事故的预测结果符合真实情况;交通流由自由流转向拥堵过程中追尾事故风险最大,拥堵中运动波的传播会增加追尾事故风险。

To predict the real-time collision risk for rear-end crashes at freeway recurrent bottlenecks, in this study, the features of trajectories of individual vehicles when encountering kinematic waves were investigated, the collision conditions were analyzed, aggregate analysis of the trajectories of multiple vehicles was conducted. Then a model to predict the collision risks was developed using aggregated traffic flow data. Results show that the rear-end collision probability during propagation of kinematic waves was significantly related to the collision risk index and standard deviation of upstream occupancy. The model was used to predict the collisions in the bottleneck section of a real-world freeway; the predictions were consistent with observations. The highest collision risk occurs when traffic changes from free-flow to congested states. The propagation of kinematic waves in congested traffic also increases the rear-end collision risk.