【作者】 赵娜乐；

【导师】 于雷；

【作者基本信息】 北京交通大学 ， 交通运输规划与管理， 2010， 博士

【摘要】 摘要：随着大城市交通网络建设的发展,快速路已成为城市交通大动脉,在城市交通系统中发挥着不可替代的作用。对快速路构建动态交通模型并进行动态交通态势评估,是对大城市快速路系统乃至全网的交通管理及优化方案实施的重要基础。作为动态交通模型的路网供给能力的标定参数,路网中每条快速路路段的自由流速度、通行能力、临界速度及阻塞密度四个交通流特征参数的合理设定是准确进行动态交通态势评估的关键。利用常规方法估计交通流特征参数,需要获得快速路路段的24小时交通流数据,并选取适用于快速路交通流特性描述的交通流模型进行曲线拟合。然而,由于快速路网中大部分快速路路段未设检测器,且部分已设检测器路段无法获得各交通状态下的交通流数据样本点,使得依赖于交通流数据的传统交通流特征参数估计方法难以应用。此外,由于出入口及公交设施密集等因素,快速路的交通流特性更为复杂,基于高速公路数据建立的交通流模型对快速路的适用性也有待研究。在此背景下,本文提出了基于物理属性的城市快速路交通流特征参数模型。在对快速路交通流特性定量分析的基础上,提出利用快速路物理特征因素直接推导交通流特征参数的思路,面向应用建立基于物理属性的快速路交通流特征参数模型。主要研究成果包括：1、对快速路交通流特性进行定量研究,提出快速路时空平均速度转化公式及占有率-密度转化公式；采用适用于快速路的Van Aerde四参数单一结构模型,在对快速路断面交通流特征参数综合分析的基础上,对比分析基本路段、出入口附近路段及公交车站附近路段的确定型交通流特征参数特点；提出利用边界线分析通行能力区域筛选算法进行通行能力分布分析,经实测数据分析表明,快速路通行能力分布服从Gamma及Lognormal分布。2、提出基于物理属性的快速路交通流特征参数模型建模框架,以系统性实现直接建立快速路物理特征因素和各交通流特征参数的函数关系的基本思路,其中快速路断面特征矩阵聚类分析及快速路断面交通流特征参数模型构建为核心模块。3、建立快速路断面特征矩阵聚类模型,以实现物理-交通流特征强相关快速路断面样本提取。子成果包括：(1)提出出入口及公交设施等快速路基本物理特征；(2)建立基于模拟退火遗传算法(SAGA)优化聚类数的K-Means聚类算法的快速路断面特征矩阵聚类核心模型；(3)将物理特征数据及交通流特征数据视为二源特征数据,提出快速路断面二源特征数据类型匹配模型。4、构建快速路断面交通流特征参数模型,子成果包括：(1)基于物理特征基本因素,提出含出入口综合指标及公交综合指标的物理特征因素；(2)基于特征强相关快速路断面样本,建立物理特征因素与各交通流特征参数的多元交通流特征参数模型,并实现参数估计及模型验证。5、以北京西三环快速路交通走廊动态交通模型为基础平台,对所建立的快速路交通流特征参数模型进行测试。结果表明,在出入口及公交设施密度大且缺乏有效实测交通流数据情况下,较之传统的采用统一交通流特征参数的方法,利用所建立的交通流特征参数模型进行动态交通模型的交通流特征参数标定,能更准确地捕捉快速路交通流特性。更多还原

【Abstract】 ABSTRACT:With the development of road networks in megacities, expressways have become the urban traffic artery and played an irreplaceable role in urban transportation systems. Developing the dynamic traffic model and assessing the dynamic traffic states is the important foundation for the implementation of traffic management and optimization for not only expressway systems but the entire road network of a megacity as well. The determination of the calibration parameters of the road network supply capacity in the dynamic traffic model, e.g. the traffic flow characteristic parameters of each road section of expressways in the road network including the free flow speed, capacity, critical speed and jam density, is the key to the accurate assessment of dynamic traffic states. The traditional method in estimating traffic flow characteristic parameters needs to obtain 24h traffic flow data of sections, select traffic stream model which is suitable to describe the traffic flow characteristics of expressways, and conduct curve fitting. However, since most of road sections of expressways do not have detectors, and the traffic flow data sample could not be obtained in each traffic state in several detected road sections, the traditional method in estimating traffic flow characteristic parameters that has heavily relied on traffic flow data is hardly applied in practice. Moreover, since traffic characteristics of expressways are more complex than those of other classes of roads because of the high densities of ramps and bus stops, the applicability of existing traffic stream models, which have been developed primarily for freeways, should be examined.In this context, the research in this dissertation strives to develop traffic flow characteristic parameter models of urban expressways based on physical properties. After a quantitative analysis of traffic characteristics of expressways, it presents a new approach to deriving traffic flow characteristic parameters directly from physical characteristic parameters of expressways. As a result, the application-oriented traffic flow characteristic parameter models of expressways based on physical properties are developed. The research in the dissertation consists of the following accomplishments:1. Quantitative analyses of traffic flow characteristics of expressways are conducted, and formulas that transform the time mean speed to the space means speed and the occupancy to the density are proposed. Using the Van Aerde single-regime model, which is considered suitable for expressways, the dissertation analyzed and compared deterministic traffic flow characteristic parameters of the basic section, the section near ramps and the section near bus stops based on the comprehensive analysis of traffic flow characteristic parameters of expressways,. The algorithm for the selection of the range of capacities based on the Bound-line analysis is proposed and the stochastic distribution analysis of capacity is conducted. The analysis of the field data indicates that the capacity of expressways follows the Gamma distribution and Lognormal distribution.2. A framework for building traffic flow characteristic parameter models for expressways based on physical properties is established in order to identify direct functions between physical characteristic parameters and each of traffic flow characteristic parameters. The framework includes two essential modules, the character matrix cluster analysis of sections and the development of the traffic flow characteristic parameter models on expressways.3. The character matrix clustering model for sections on the expressway is established, attempting to select the sectional samples with a high correlation between physical and traffic flow characteristics. The relevant achievements include:(1) basic physical characteristic parameters of expressways that consider both series of ramps and transit facilities are identified; (2) based on the SAGA-K Means cluster algorithm, the character matrix cluster model of the section on the expressway is established; and (3) a cluster matching model for dual sources of data for expressways is developed by considering the physical characteristic data and traffic flow characteristic data as two sources of data,4. The traffic flow characteristic parameter models for the section on the expressway based on physical properties are developed. The relevant achievements include:(1) on the basis of basic physical characteristic parameters, the physical characteristic parameters incorporating the ramp comprehensive index and the transit comprehensive index are proposed; and (2) using samples with a high correlation, the multivariate models of physical characteristic parameters and traffic flow characteristic parameters are developed, and the parameter estimation and model validation are conducted.5. Using the dynamic traffic model of the western 3rd ring-road expressway corridor of Beijing as the basic platform, the proposed traffic flow characteristic parameter models are tested. The results indicate that, with the high densities of ramps and transit facilities and in the absence of valid observed traffic flow data, calibrating the dynamic traffic model by the proposed traffic stream model could better capture the traffic flow characteristics of expressways than could the traditional method that uses the uniform traffic flow characteristic parameters.更多还原