基于智能驾驶员模型的危险跟驰行为特征分析

Characteristics Analysis of Dangerous Car-following Behavior Based on IDM Model

为提高危险跟驰行为研究的效率和可靠度,创新研究基于智能驾驶员模型(IDM)模型参数的危险跟驰行为定义方法。首先,以NGSIM自然车辆数据集为基础提取跟驰轨迹数据,并通过遗传算法标定IDM模型参数,挖掘驾驶员跟驰行为特征;其次,在分析跟驰行为特征指标分布规律的基础上,考虑指标间的相关性,确定各等级危险跟驰行为指标阈值;最后,设计车辆跟驰的仿真试验,选择6个指标对不同级别危险场景下的交通运行仿真结果进行评价。结果表明,危险驾驶员跟车间距更小,间距波动系数较大,并且会干扰交通整体运行。

To improve the efficiency and reliability of the research on dangerous car-following behavior,the dangerous car-following behavior was defined based on IDM model parameters in this paper.First,extracting the car-following trajectory data based on the NGSIM natural vehicle data set,and calibrating the IDM model parameters through genetic algorithm to mine the driver’s car-following behavior characteristics.Then,based on the distribution characteristic of car-following behavior indicators,the threshold of each level of dangerous car-following behavior index was determined considering the correlation between indicators.Finally,a car-following simulation test was designed,and six indicators were selected to evaluate the performance of the traffic network in different levels of dangerous scenarios.The results show that the head spacing between dangerous drivers is smaller,and the distance fluctuation coefficient is larger.Meanwhile,when there is a large percentage of dangerous drivers,the overall network performance will be worse.