【作者】 刘伟；

【导师】 夏井新；

【作者基本信息】 东南大学 ， 交通运输工程， 2019， 硕士

【摘要】 区域门限控制作为一种新兴的区域控制方法为缓解区域交通拥堵提供了新的思路,其中子区划分和承载力估计是区域门限控制的重要支撑。现有门限控制子区划分方案存在对子区内部交叉口控制协调性考虑不足、子区划分连通性较差等问题。本研究以车辆出行轨迹数据为基础,以社团划分、谱聚类等方法为手段,研究包含基于交叉口间车辆出行轨迹关联性的初步控制子区划分、基于路段车辆密度分布的门限控制子区划分的双层控制子区划分方法,并对门限控制子区承载力进行研究。具体研究内容如下。本文初步控制子区划分方法基于车辆出行轨迹信息和社团划分算法对路网进行初步划分,使出行轨迹或出行需求集中的区域形成一个相对独立的初步控制子区,以提升子区内交叉口信号控制的协调性,为后续的门限控制子区划分提供一个初步的划分方案。本方法首先基于车辆出行轨迹分析路网所有交叉口对之间的关联性,构建以交叉口为顶点、以交叉口对关联性为边权的全联通无向加权图,不断对无向加权图的关联边进行筛选并利用Newman快速算法进行子区划分,在不同无向加权图关联边筛选比例下选择子区连通性最好的方案作为最优初步控制子区划分方案。在初步控制子区划分方案的基础上,将初步控制子区进一步精细划分成多个门限控制子区,使每个子区内部路段车辆密度相近,子区间路段车辆密度均值相差较大,以便后期更加准确地进行区域承载力估计,为区域门限控制提供合理的控制边界。本部分首先利用车辆出行轨迹数据估计各路段的车辆密度,然后基于“local scale”思想计算路段之间的相似度并构建相似度矩阵,在此基础上利用多路谱聚类算法将路网划分为多个初始子区并迭代融合确定较优的划分方案,最后寻找不同初始划分子区数下的最佳划分方案作为最终的区域门限控制子区划分方案。通过将本方法与二路谱聚类迭代划分-合并方法进行对比,结果显示本方法具有更好的普适性和划分效果,为后续区域承载力估计提供了较为合理的区域划分方案。区域路网承载力反映了路网在稳定交通状态下所能承受的最大累积车辆数,是城市交通管控的重要支撑。传统计算承载力的方法存在结构复杂、数据难以获取、理论假设难以验证的缺陷,本文基于近年来新兴的宏观基本图理论来分析门限控制子区的承载力,具有数据获取容易、模型关系简单、实用性高的突出优点。在区域门限控制子区划分方案的基础上,本部分首先基于车辆出行轨迹数据构建门限控制子区的MFD,并对子区一天各时段、一周各天早晚高峰的MFD进行了分析;为了对MFD的变量建立定量关系,本部分基于H-S主曲线算法对MFD散点进行去噪,然后对比多种拟合函数对主曲线质心进行拟合,确定有理函数更加符合MFD的变化趋势,并基于此对子区工作日、周末早晚高峰承载力进行了估计。更多还原

【Abstract】 As an emerging regional control method,regional gating control provides a new idea for mitigating regional traffic congestion.Network partition and carrying capacity estimation are important support for regional gating control.The existing network partition for gating control has problems such as insufficient coordination of internal intersections in subzones,and does not organically combine purpose of various control subzones.Based on the vehicle travel trajectory data,this study uses the method of community division and spectral clustering as the means to study the preliminary control sub-area based on the correlation of the vehicle’s travel trajectory at the intersection,and the gating control sub-area based on the vehicle density distribution of the road segment,and the carrying capacity of gating control subzone is studied.The specific research content is as follows.The preliminary control sub-area division method based on the vehicle travel trajectory information and community division algorithm preliminarily divides the road network,so that the travel trajectory or the area where the travel demand is concentrated forms a relatively independent preliminary control sub-area to enhance the intersection in the sub-area.The preliminary control subzone division method provides a preliminary division scheme for the subsequent division of the gating control subzone.The method firstly analyzes the correlation between all pairs of intersections on the road network based on the vehicle travel trajectory,and constructs a non-directional weighted graph of the all-connectivity with the intersection as the vertex.The correlation edge is selected repeatedly and the Newman fast algorithm is used to divide the subzone.The scheme with the best subzone connectivity is selected as the optimal preliminary control subzone partitioning scheme under different undirected weighted graph correlation edge screening ratios.On the basis of the preliminary control sub-area division scheme,the preliminary control sub-area is further finely divided into multiple gating control sub-areas,so that the density of vehicles in the internal sections of each sub-area is similar,and the mean value of vehicle density in sub-section sections is relatively large,so that later Accurately carry out regional carrying capacity estimation,and provide more reliable regional road network critical state information for threshold control.This method first uses the vehicle travel trajectory data to estimate the vehicle density of each road segment,and then calculates the similarity between the road segments based on the "local scale" idea and constructs a similarity matrix.On this basis,the multi-way spectral clustering algorithm is used to divide the road network into multiple initial subzones and subzone iteratively merge to determine the better partitioning scheme.Finally,the best partitioning scheme under different initial subzones is used as the final regional gating control subzone partitioning scheme.By comparing the method with the twoway spectral clustering iterative partitioning-merging method,the results show that the proposed method has better universality and partitioning effect,which provides a better regional partitioning scheme for subsequent regional carrying capacity estimation.The regional road network carrying capacity reflects the maximum cumulative number of vehicles that the road network can withstand under stable traffic conditions,and is an important support for urban traffic control.The traditional method of calculating the carrying capacity has the defects of complex structure,difficult data acquisition,and difficult to verify theoretical assumptions.Based on the emerging MFD theory in recent years,this paper analyzes the carrying capacity of the gating control subzone,which has the outstanding advantages of easy data acquisition,simple model relationship and high practicability.Based on the regional gating control subzone division scheme,this part first constructs the MFD of the gating control subzone based on the vehicle travel trajectory data,and analyzes the MFD of the subzone at each time of day,the morning and evening peaks of each day of the week.In order to establish a quantitative relationship between the variables of MFD,this part denoises the MFD scatter based on the H-S principal curve algorithm,and then fits the centroid of the main curve by comparing various fitting functions to determine that the rational function is more in line with the change trend of MFD.Based on this,the peak carrying capacity of the subzone workday and weekend morning and evening was estimated.更多还原