中文摘要：

根据交通拥挤状态下交通流速度与密度一致性变化的特点,分析了拥挤交通流的平均车间时距为定值的原因,并结合最小安全间距约束提出了交通拥挤状态下的速度-密度关系模型。研究了驾驶人的平均反应时间和交通拥挤状态下的最小车间时距的关系,对速度-密度关系模型的反应时间进行参数标定。分析了不同车辆长度、阻塞停车间距和反应时间下的速度-密度关系,利用提出的速度-密度关系模型、Greenshields模型、Greenberg模型、Underwood模型、Northwest模型、Edie模型对美国US-101、I-80两条高速公路的交通数据进行拟合,得到了拟合结果和绝对误差。分析结果表明：提出的速度-密度关系模型能够从理论上解释交通拥挤状态下速度与密度的变化关系和速度-密度数据的离散现象;和其他模型相比,提出的速度-密度关系模型在拟合2条高速公路交通数据时的绝对误差最小,分别为4.91、7.50veh·km^-1。基于最小安全间距约束的速度-密度模型刻画了拥挤交通流的本质特征,且对现实数据能够取得更好的拟合效果。

英文摘要：

Based on the conforming change characteristic of speed and density for traffic flow under traffic congestion state, the reason that the average headway-distance of congested traffic flow was a fixed value was analyzed, and the speed-density relation model under traffic congestion state was put out combined with minimum safety distance constraint. The relation between the average reaction time of driver and the minimum headway-distance under traffic congestion state was discussed, and the parameter calibration of reaction time for speed-density relation model was carried out. The speed-density relations with different vehicle lengths, blocking parking distances and reaction times were analyzed, the traffic data of US-101 Highway and 1-80 Highway in America were fitted by using the proposed speed-density relation model, Greenshields model, Greenberg model, Underwood model, Northwest model and Edie model respectively, and thefitting results and absolute errors were obtained. Analysis result shows that by using the proposed speed-density relation model, the change relation of speed and density under traffic congestion state can be explained in theory, and the discrete phenomena of speed-density data can be described. Compared with the other models, the absolute errors of fitting results for US-101 Highway and 1-80 Highway are minimum by using the proposed speed-density relation model, and are 4.91, 7.50 veh · km^-1 respectively. So the proposed speed-density relation model based on minimum safety distance constrain reveals the substantive characteristic of congested traffic flow, and the fitting effect on actual data is better. 1 tab, 12 figs, 24 refs.