波阵面视角下的车辆拥堵分析

基于视频数据，分析了交通事故下车辆运动的复杂性和队长变化的本质，借鉴车流波动理论、车辆跟驰理论、流体力学理论等，提出利用波阵面的传播轨迹描述排队过程，并给出了排队长度动态模型VQLM.同时，为减少视频图像与实际距离的转换误差，提出以标准当量车为参考，多次迭代还原真实距离的标定方法.研究结果表明：VQLM模型较基于单一理论计算队长的方法更精确、实用.模型验证结果：平均队长误差为-3.029%，其宏观表现为2辆当量车10 m内，延时误差为7 s（信号周期为60 s）.VQLM模型主要考虑了司机驾驶的主观性、道路通行强度、交通波节点函数等；数据还原标定方法简单实用，比传统比例尺算法更为精确.

Vehicle congestion from the perspective of wave front

This paper studied the complexity of the vehicle motion and the essence of the dynamic vehicle queue length based on the traffic video data. According to the traffic flow wave theory, vehicle-following theory, hydromechanics theory, etc, the method based on the propagation of wave surface to describe the queuing process was applied, and a queue length calculation model named Vehicle Queuing Length dynamic Model (VQLM) was proposed. To reduce the error during the transformation from video to reality, we introduced a calibration method which adopted passenger car unit(PCU) to recover the reality by-recursire iteration. Preliminary results show that the VQLM is more accurate and pragmatic in calculating the vehicle queue length than other methods based on single theories. According to the experiments, the average vehicle queuing length error is reduced to -3.029% i.e., when the distance error is within 10 m (2 PCU which are both 5 m long ); The time error is 7 s (signal cycle length is 60 s).Moreover, VQLM accounts more about the subjectivity of the drivers, traffic intensity of the road and node-set function of traffic wave, etc.