基于基本图和三相交通流理论的离散建模方法研究
发布时间：2018-08-25 14:21
【摘要】：科学解决交通问题的首要任务是深入研究交通流的演化机理,掌握交通流的运行规律。现阶段的交通流理论已发展成以传统基本图方法与新兴三相交通流理论为核心的科学体系。基本图方法假设稳态交通流的流量和密度具有一一对应的函数关系,并将交通流分成自由流和拥挤流两类。三相交通流理论进一步将拥挤流细分成宽运动堵塞和同步流,并假设稳态同步流在流量密度平面上呈现二维分布。合理解决基本图方法和三相交通流理论的争议问题一直是交通流理论研究的热点和难点。遵循相互包容、协同发展的科学原则,本文立足于基本图方法,解决了三相交通流理论对基本图方法的部分批判问题,在深入研究格子流体力学和元胞自动机两类离散模型的基础上提出了更加符合实测现象的模型,通过理论分析和数值模拟研究了交通流的各种非线性现象,探讨了不同类型模型之间的内在联系。研究工作如下： 1)在Nagatani格子流体力学模型基础上,提出了考虑前后格子密度差效应的格子模型(DDLM模型)。通过线性稳定性分析得到了DDLM模型的中性稳定性曲线,并发现考虑密度差效应可以提高交通流的稳定性。通过非线性稳定性分析得到了稳定区域、亚稳定区域和不稳定区域存在的以Burgers方程、KdV方程和mKdV方程描述的三角激波、孤立波和扭结波。随后提出了匝道系统的格子流体力学模型,并设计了随机型和确定型匝道两种数值模拟方法。随机型匝道的模拟结果和基本图方法的实测结果一致,确定型匝道的模拟结果发现了四种新的组合拥挤模式,发现DDLM模型的堵塞可以产生于同步流,因此三相交通流理论对基本图方法模型堵塞产生于自由流的批判不适用于DDLM模型。 2)在BL刹车灯模型的基础上提出了期望时间间距刹车灯模型(DTGBLM模型)。DTGBLM模型不仅可以产生符合实测的同步流以及三相交通流理论的扩展同步流模式和全面模式,而且过大减速度现象出现的概率可忽略不计,因此DTGBLM模型解决了三相交通流理论对BL模型的批判问题。其次在FMCD和BLDAD刹车灯模型的基础上构建了提前慢化刹车灯模型(ARBLM模型)。ARBLM模型需借助外界扰动才能形成同步流,这和实测通行能力陡降现象借助瓶颈产生的机理一致,因此ARBLM模型不仅可以解释通行能力陡降现象的产生机理,而且比三相交通流理论的FMCD模型更加符合实际。 3)从传统的多车期望效应跟驰模型思想出发,将多车平均间距作为车辆的期望速度引入随机慢化过程,提出了平均车辆间距元胞自动机模型(ASGM模型)。ASGM模型不仅可以模拟自由流到同步流以及同步流到宽运动堵塞的一阶相变,而且可以模拟三相交通流理论的各种实测拥挤模式。由于ASGM模型和大多数三相交通流理论模型的模拟结果没有本质区别,稳态拥挤交通流在流量密度平面上呈现二维分布不是模型符合实测的基本条件。深入分析多车期望效应跟驰模型,提出了扩展平均车辆间距元胞自动机模型(EASGM模型)。EASGM模型能够模拟宽运动堵塞的同步出流,并且证实了车辆期望时间间距、慢启动的随机慢化概率和安全控制参数都可影响宽运动堵塞的出流状态。ASGM模型和EASGM模型不仅建立了连续的跟驰模型和离散的元胞自动机模型之间的联系,而且解决了三相交通流理论批判基本图方法不能模拟宽运动堵塞同步出流的问题。 4)提出了三阶段标定法,并对上述新模型和几个不同类型、不同理论体系的模型进行了标定。三阶段标定法首先挑出可以根据缺省值和实测数据直接给定的参数,然后采用试差法或启发式算法标定模型的其他参数,最后从第一步中挑选出能够直接影响车速的参数,并再次采用试差法或启发式算法对其进行标定。标定结果表明新模型的标定误差均小于原模型;三相交通流理论模型和基本图方法模型之间误差相当,没有优劣之分。
[Abstract]:The primary task of solving traffic problems scientifically is to study the evolution mechanism of traffic flow and grasp the operation law of traffic flow. At present, traffic flow theory has developed into a scientific system centered on the traditional basic graph method and the emerging three-phase traffic flow theory. The three-phase traffic flow theory further subdivides the congestion flow into wide motion blockage and synchronous flow, and assumes that the steady-state synchronous flow presents a two-dimensional distribution on the flow density plane. Following the scientific principle of mutual inclusion and coordinated development, this paper solves the partial criticism of the basic graph method in the three-phase traffic flow theory based on the basic graph method. On the basis of the in-depth study of the lattice fluid mechanics and cellular automata discrete models, a model more in line with the measured phenomena is proposed. By theoretical analysis and numerical simulation, various nonlinear phenomena of traffic flow are studied, and the internal relations among different types of models are discussed.
1) On the basis of Nagatani lattice hydrodynamic model, a lattice model (DDLM model) considering the effect of lattice density difference is proposed. The neutral stability curve of DDLM model is obtained by linear stability analysis, and it is found that the stability of traffic flow can be improved by considering the effect of density difference. Triangular shock, solitary wave and kink wave are described by Burgers equation, KdV equation and mKdV equation in the region, metastable region and unstable region. Lattice hydrodynamic model of ramp system is proposed, and two numerical simulation methods of random and deterministic ramps are designed. Simulation results and basic graph method of random ramps are given. The simulation results of deterministic ramps show that there are four new combined congestion modes. It is found that the congestion of DDLM model can occur in synchronous flow. Therefore, the critique of three-phase traffic flow theory that the congestion of basic graph method model originates from free flow is not applicable to DDLM model.
2) Based on the BL brake lamp model, the expected time interval brake lamp model (DTGBLM model) is proposed. The DTGBLM model can not only generate the extended synchronous flow model and the comprehensive flow model which conform to the measured synchronous flow and the three-phase traffic flow theory, but also neglect the probability of excessive deceleration. Therefore, the DTGBLM model solves the problem. Secondly, the ARBLM model is built on the basis of FMCD and BLDAD brake lamp model. The ARBLM model needs external disturbance to form synchronous flow, which is consistent with the mechanism of the phenomenon of abrupt capacity drop with the help of bottleneck, so the ARBLM model can not only be used. The mechanism of the abrupt drop of traffic capacity is explained, and it is more practical than the FMCD model of three-phase traffic flow theory.
3) Based on the traditional idea of multi-vehicle expected effect car-following model, the average vehicle spacing is introduced into the stochastic slowing process as the expected vehicle speed, and the average vehicle spacing cellular automata model (ASGM model) is proposed. Because there is no essential difference between the simulation results of ASGM model and most of the three-phase traffic flow theoretical models, the two-dimensional distribution of steady-state congestion traffic flow on the flow density plane is not the basic condition for the model to meet the actual measurement. Extended Average Vehicle Spacing Cellular Automata Model (EASGM model) is proposed. EASGM model can simulate the synchronous outflow with wide motion jam, and it is proved that the expected vehicle spacing, stochastic slowing down probability of slow start and safety control parameters can affect the outflow state with wide motion jam. The connection between the car-following model and the discrete cellular automata model solves the problem that the critical basic graph method of three-phase traffic flow theory can not simulate the synchronous outflow with wide motion blockage.
4) Propose a three-stage calibration method, and calibrate the new model and several models of different types and different theoretical systems. The three-stage calibration method first selects parameters that can be given directly according to default values and measured data, then calibrates other parameters of the model by trial-error method or heuristic algorithm, and finally selects the parameters from the first step. The calibration results show that the calibration error of the new model is smaller than the original model, and the error between the three-phase traffic flow theoretical model and the basic graph method model is equal, and there is no difference between the advantages and disadvantages.
【学位授予单位】：北京交通大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：U491.112
本文编号：2203137
[Abstract]:The primary task of solving traffic problems scientifically is to study the evolution mechanism of traffic flow and grasp the operation law of traffic flow. At present, traffic flow theory has developed into a scientific system centered on the traditional basic graph method and the emerging three-phase traffic flow theory. The three-phase traffic flow theory further subdivides the congestion flow into wide motion blockage and synchronous flow, and assumes that the steady-state synchronous flow presents a two-dimensional distribution on the flow density plane. Following the scientific principle of mutual inclusion and coordinated development, this paper solves the partial criticism of the basic graph method in the three-phase traffic flow theory based on the basic graph method. On the basis of the in-depth study of the lattice fluid mechanics and cellular automata discrete models, a model more in line with the measured phenomena is proposed. By theoretical analysis and numerical simulation, various nonlinear phenomena of traffic flow are studied, and the internal relations among different types of models are discussed.
1) On the basis of Nagatani lattice hydrodynamic model, a lattice model (DDLM model) considering the effect of lattice density difference is proposed. The neutral stability curve of DDLM model is obtained by linear stability analysis, and it is found that the stability of traffic flow can be improved by considering the effect of density difference. Triangular shock, solitary wave and kink wave are described by Burgers equation, KdV equation and mKdV equation in the region, metastable region and unstable region. Lattice hydrodynamic model of ramp system is proposed, and two numerical simulation methods of random and deterministic ramps are designed. Simulation results and basic graph method of random ramps are given. The simulation results of deterministic ramps show that there are four new combined congestion modes. It is found that the congestion of DDLM model can occur in synchronous flow. Therefore, the critique of three-phase traffic flow theory that the congestion of basic graph method model originates from free flow is not applicable to DDLM model.
2) Based on the BL brake lamp model, the expected time interval brake lamp model (DTGBLM model) is proposed. The DTGBLM model can not only generate the extended synchronous flow model and the comprehensive flow model which conform to the measured synchronous flow and the three-phase traffic flow theory, but also neglect the probability of excessive deceleration. Therefore, the DTGBLM model solves the problem. Secondly, the ARBLM model is built on the basis of FMCD and BLDAD brake lamp model. The ARBLM model needs external disturbance to form synchronous flow, which is consistent with the mechanism of the phenomenon of abrupt capacity drop with the help of bottleneck, so the ARBLM model can not only be used. The mechanism of the abrupt drop of traffic capacity is explained, and it is more practical than the FMCD model of three-phase traffic flow theory.
3) Based on the traditional idea of multi-vehicle expected effect car-following model, the average vehicle spacing is introduced into the stochastic slowing process as the expected vehicle speed, and the average vehicle spacing cellular automata model (ASGM model) is proposed. Because there is no essential difference between the simulation results of ASGM model and most of the three-phase traffic flow theoretical models, the two-dimensional distribution of steady-state congestion traffic flow on the flow density plane is not the basic condition for the model to meet the actual measurement. Extended Average Vehicle Spacing Cellular Automata Model (EASGM model) is proposed. EASGM model can simulate the synchronous outflow with wide motion jam, and it is proved that the expected vehicle spacing, stochastic slowing down probability of slow start and safety control parameters can affect the outflow state with wide motion jam. The connection between the car-following model and the discrete cellular automata model solves the problem that the critical basic graph method of three-phase traffic flow theory can not simulate the synchronous outflow with wide motion blockage.
4) Propose a three-stage calibration method, and calibrate the new model and several models of different types and different theoretical systems. The three-stage calibration method first selects parameters that can be given directly according to default values and measured data, then calibrates other parameters of the model by trial-error method or heuristic algorithm, and finally selects the parameters from the first step. The calibration results show that the calibration error of the new model is smaller than the original model, and the error between the three-phase traffic flow theoretical model and the basic graph method model is equal, and there is no difference between the advantages and disadvantages.
【学位授予单位】：北京交通大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：U491.112
【参考文献】
相关期刊论文 前7条
1 陈建阳;交通流微观模型与宏观模型的统一[J];同济大学学报(自然科学版);1997年01期
2 雷丽,薛郁,戴世强;交通流的一维元胞自动机敏感驾驶模型[J];物理学报;2003年09期
3 薛郁;随机计及相对速度的交通流跟驰模型[J];物理学报;2003年11期
4 邝华,孔令江,刘慕仁;多速混合车辆单车道元胞自动机交通流模型的研究[J];物理学报;2004年09期
5 关伟;何蜀燕;马继辉;;交通流现象与模型评述[J];交通运输系统工程与信息;2012年03期
6 董力耘,薛郁,戴世强;基于跟车思想的一维元胞自动机交通流模型[J];应用数学和力学;2002年04期
7 戴世强,冯苏苇,顾国庆;交通流动力学:它的内容、方法和意义[J];自然杂志;1997年04期
，本文编号：2203137
本文链接：https://www.wllwen.com/kejilunwen/jiaotonggongchenglunwen/2203137.html
