论文部分内容阅读
车辆间隙定义为到达路段某特定点相邻车辆的间隔.本文研究了非双车道异质混合交通情形下的车辆间隙模型,此类交通流在诸如印度等一些发展中国家普遍存在.其特征是具有大量“零间隙”的情况,这主要由给定车道宽度下车辆同时到达的情形所致.其另一特征是较大车头时距带来的尾区数据量偏大.然而,一些学者运用轻尾分布对车辆间隙进行建模并得到了满意的结果.原因主要是:(1)尾区数据融入单个面元;(2)分布拟合时卡方检验等方法具有一定的局限性.此外,一些学者还建议对同一范围的交通流数据采用其他一些分布进行拟合,产生了分布选择的分歧.同时,可表征任何分布拟合状况的面元在间隙模型中的作用逐渐减小.因此,本文就以上问题对现有间隙模型的研究进行进一步和标准化的分析.本文还分析了两种具有较好尾模型特性的广义Pareto分布(GP)、广义极值分布(GEV)、其他几种传统分布对550 vph至4 100 vph流量范围内车辆间隙建模,且使用基于区域和基于距离检验的方法进行分布拟合检验.结果表明,无论使用哪种检验方法,广义Pareto分布对于大于1 500 vph流量的间隙数据在整体和尾区均能得到较好的拟合效果.广义极值分布在使用基于区域的检验方法时对于大于1 500 vph流量的间隙数据产生较好的拟合效果.
Vehicle clearance is defined as the distance between adjacent vehicles arriving at a particular point in the road segment.This paper studies the vehicle clearance model in the case of non-dual-lane heterogeneous mixed traffic, which is common in some developing countries such as India and is characterized by This is mainly due to the simultaneous arrival of vehicles at a given lane width, as is the case with a large number of “zero clearances.” Another feature of this is the large amount of data in the tail zone due to larger headway. However, The authors use the tail distribution to model the vehicle clearance and obtain satisfactory results.The main reasons are as follows: (1) the tail zone data is integrated into a single bin; (2) the chi-square test and other methods have some limitations In addition, some scholars also proposed to adopt some other distributions to fit the same range of traffic flow data, resulting in disagreement of distribution choices.At the same time, the role of bins that can characterize the distribution of any distribution in the gap model is gradually reduced Therefore, this paper further and standardizes the research on the existing gap model based on the above problems.This paper also analyzes two generalized Pareto distributions (GPs) with good tail model properties, generalized pole Distribution (GEV), several other conventional distribution models for vehicle clearances in the 550 vph to 4 100 vph flow range, and distribution-fitting tests using region-based and distance-based tests.The results show that regardless of the test method used , The generalized Pareto distribution can obtain a better fitting effect on the whole and the tail region for the gap data larger than 1 500 vph.Generalized extreme value distribution is generated for the gap data greater than 1 500 vph when using the region-based method Better fitting effect.