Abstract: Car-following behavior is one of the most fundamental behaviors in micro traffic flow. The NGSIM trajectory data was employed to analyze the micro characteristics of the car-following behavior, and the influencing factors and causes of car-following behavior from safe state to dangerous state were analyzed. Time to collision (TTC) was selected as the safety index. The number of TTCs in different states was used as the dependent variable, and the micro characteristics such as speed, acceleration, and acceleration difference were used as independent variables, and four thresholds were set for regression analysis. Results show that the biggest factor affecting the following behavior from safe state to dangerous state is the speed difference between the front and rear vehicles, followed by the distance between the front and rear vehicles. However, when there are a large number of following behaviors, the distance between the front and rear vehicles has a greater impact, and the micro characteristics of the rear vehicle have more impact on this than the front vehicle. In order to keep the car-following behavior within a safe state, the speed, acceleration, acceleration difference of the leading cars, and the distance between the leading and following cars should be increased. At the same time, the acceleration, acceleration difference of the following cars, and the speed difference between the leading and following cars should be reduced. Considering the continuity of traffic flow, on the whole, the speed difference between the leading and following cars should be narrowed, sudden acceleration fluctuations should be avoided, and more attentions should be paid to maintain the proper distance between the leading and following cars and strengthen the control of the following cars.

Key words: traffic engineering; NGSIM; car-following behavior; time to collision; regression analysis; safety evaluation

摘要： 跟驰行为是微观交通流中最基本的交通行为之一，选用NGSIM轨迹数据对车辆跟驰行为的微观特性进行研究，分析跟驰行为由安全状态变为危险状态的影响因素及原因。选取碰撞时间（time to collision, TTC）为安全指标，将TTC处于不同状态下的帧数作为因变量，将速度、加速度、加速度差分等微观特性变量作为自变量，并设定4种阈值回归分析。结果表明：对跟驰行为由安全状态变为危险状态，影响最大的因素为前后车速度差，前后车平均车头间距次之，但当存在大量跟驰行为时，前后车平均车头间距的影响更大，后车的微观特征对此产生的影响多于前车；为使跟驰行为保持在安全状态，应增加前车的速度、加速度、加速度差分和前后车平均车头间距，同时应减小后车的加速度、加速度差分和前后车速度差值；考虑到车流连续性，整体而言应使前后车保持相近的速度，避免突然加速或减速，且应保持适当的车头间距，并加强后车的管控。

关键词: 交通工程；NGSIM；跟驰行为；碰撞时间；回归分析；安全评价