基于Poisson过程的交通冲突预测模型研究

Stochastic Process-based Traffic Conflict Predicting Model

交通冲突预测是利用交通流运行特征,预测出可能发生交通冲突的次数,从而为道路交叉口等交通交汇处的交通安全评价提供一种较为可靠和可行的方法。影响车辆发生冲突的因素包括车辆在冲突点的暴露时间、车辆到达潜在冲突点的特性以及车辆间的车头时距等。在对交通冲突点的冲突过程分析的基础上,给出了交通冲突次数的定义,包括大于2辆车连续通过冲突点的交通状况.应用Poisson随机过程理论,建立潜在冲突点机动车间交通冲突次数的预测模型。预测结果的对比分析研究表明,基于随机过程的交通冲突预测模型的预测值与实际观测值最接近。

Traffic conflict predicting is defined as predicting the times of the possible conflicts by means of analyzing the characteristic of traffic flow,and providing a dependable and feasible method for traffic safety evaluation of transport interchange,such as intersections.The factors which influence the conflicts of vehicles refer to the exposure time at the conflict point,the arriving characteristics during the exposure time,and the time headway among the vehicles,etc.By analyzing the conflict process of traffic conflict point,this paper presents a definition for the times of traffic conflicts.By applying the theory of Stochastic Process to it,the paper constructs a predicting model for the times of traffic conflicts among vehicles at conflict point.The comparative analysis of the results with the other predicting models shows that the predictive accuracy has been improved.