【作者】 唐明；

【导师】 隽志才；

【作者基本信息】 吉林大学 ， 交通运输规划与管理， 2010， 博士

【摘要】 随着社会经济的快速发展和城市化进程的加速,城市机动车拥有量不断增长,交通需求急剧增加,由此带来的交通拥堵、交通事故、环境污染和能源短缺等交通相关问题已成为世界各国共同面临的难题。鼓励公共交通,重视行人交通成为当前许多国家和地区改善交通环境、实现绿色交通的一项重要管理举措。作为城市客运交通系统网络节点的综合客运枢纽,是多种交通方式、不同方向客流集散、换乘的场所,通过其功能的发挥将各种交通方式有效的衔接配合。客运交通枢纽的服务水平、换乘效率直接影响着公共交通各方式对出行者的吸引力,同时也是提高城市客运交通系统便捷性、舒适性和安全性的关键环节之一。行人流仿真是深入理解客运枢纽环境行人流复杂演化规律的有效方法,基于多智能体复杂系统建模技术建立客运枢纽行人流仿真模型以及算法,对于定量分析客运枢纽设施规划与设计方案,为提高客运枢纽服务水平及运行效率具有重要的理论价值和现实意义。现有行人流仿真模型多针对安全疏散问题而设计,且模型参数多采用国外研究标准,对于模拟复杂场景下交通活动类型多样的惯常行人流复杂演化过程,有待于从微观行人交通理论及行人智能体导航模型与算法方面完善和改进。本文首先介绍了行人流模型及相关仿真技术的国内外研究现状,并基于现场获取的实验数据,对客运枢纽行人交通特性进行分析,包括客运枢纽行人交通行为特征分析,各类设施行人流的速度-密度-流量的回归关系,使用Lyapunov指数对短时行人交通流进行了混沌特征判定。分析结果表明客运枢纽相对于道路等环境的行人期望速度偏高,各类设施的短时行人流普遍具有混沌特性,仅扶梯除外。基于交通工程学和认知心理学以及行人交通特性分析,总结了现有的行人行为理论和有关模型;根据已有的惯常三层次行人交通行为模型和建立的行人信息加工模型,完整提出了微观行人行为理论框架。在此框架下,分析了现有行为理论中欠缺的行人接受设施服务的活动过程,建立了服务活动模型;基于现场的实验数据对行人运动模型和扶梯与楼梯选择模型进行了标定工作。通过对客运枢纽功能与结构的分析,基于认知心理学中的认知Schemata理论和客体Affordance理论,基于功能对客运枢纽设施进行了分类,建立了客运枢纽设施建模;定义了客运枢纽设施的特征点和Link,分析了Link的基本特点,基于实际案例建立了客运枢纽内的设施网络拓扑模型。基于Montello导航理论、微观行人行为理论以及活动有限状态机模型,建立了客运枢纽环境下完整的行人Agent导航模型。给出了三个层次交通行为的具体算法;考虑行人在路径选择时视野受到障碍物或设施遮挡,使用当前路段行人流速度信息作为行人选择路径的主要考虑因素,提出了准静态路径选择算法。引入行人流惯常状态下行人之间的有限度竞争最优路径策略,提出了通道内的Agent动态路径目标节点更新算法,解决了通道弯道处由于Agent竞争最优路径目标节点可能导致的死锁现象问题;在行人流交织区域,由于流量较小方向上的行人流在仿真过程中容易被流量较大方向上的人群阻滞,通过研究人群子整体的判断方法,提出了行人Agent对人群障碍物的动态避障算法,解决了交织区域小流量方向上的行人Agent的死锁问题。最后针对空间连续、时间离散的行人流仿真系统的输出数据,给出了仿真行人流特征参数的统计算法,研究了行人密度分布的动态显示算法;采用K-S检验方法,使用PSSITH行人流仿真系统对行人运动模型进行了有效性验证,在Netlogo多智能体仿真平台上,对标定的社会力模型和其它社会力模型进行了仿真结果误差的对比分析;采用T检验方法对行人换乘时间仿真结果误差进行了分析;使用Lyapunov指数分析了仿真短时行人流的混沌特性。最后使用PSSITH仿真系统,进行了实例的设施瓶颈分析和交通需求量加载分析。本文的研究将丰富和拓展行人交通行为的理论框架和方法体系,提出的行人Agent导航模型及算法,可为客运枢纽等环境的行人流仿真系统提供核心算法与理论支持,从而更加深入地分析行人交通流复杂演化过程。更多还原

【Abstract】 With the quick development of society economy and continuous acceleration of urbanization process, urban vehicle population continues to increase and urban traffic demands increase sharply. Many problems about traffic that includes traffic congestion, traffic accidents, environmental pollution and energy shortage have become a universal problem. Encourage public transport and pay attention to pedestrian traffic, has been the important management strategy to achieve green traffic in many countries and regions. As the complex urban transportation network node, passenger hub is the area of vary transport mode transforming and vary direction passenger flow assembling based on joining the vary transport mode and exerting its function fully. The service level and transfer efficiency of passenger hub affects the attractive power to the passengers directly, and at the same time it is the key joint to improve the convenience, amenity and safety of urban transportation system.Constructing micro behavior models and related algorithms based on computer function and data simulation technology is one of the efficient approaches to understanding the complex evolution phenomenon of pedestrian traffic flow in passenger transfer hub. Most of current pedestrian flow simulation models and software are designed for the safety evacuation simulation and parameters of the models were always foreign standards. Its needed improved in the micro pedestrian traffic behavior theory and navigation model and algorithm of pedestrian agent for simulating the complex evolution process of pedestrian traffic flow in the complex passenger transfer hub in which the activities are varies.Firstly, pedestrian traffic flow models and related simulation technology developing status were introduced in this paper, the pedestrian traffic characteristics were analyzed in the passenger transfer hub based on the field experimental data, including the pedestrian traffic behavior, fundamental diagram of pedestrian traffic flow of vary pedestrian facilities, chaos characteristics of short-term pedestrian traffic flows. The outputs showed that the pedestrian free speed in transfer hub is higher than other environments and most pedestrian facility traffic flow have chaos characteristics, except escalators.Pedestrian behavior theory and models were analyzed based on traffic engineering and cognitive psychology theories. Pedestrian activity for receiving service of facility was analyzed, which was absent in current pedestrian behavior theory, and service activity model was constructed. A micro pedestrian behavior theory framed was proposed completely, pedestrian dynamic model and choice model of stairs and escalators were calibrated based on field data. Based on Schemata terminology and the object affordance theory in cognitive psychology field, the constructing and function of passenger transfer hub were analyzed, passenger transfer hub facilities were classified and abstract by the function and construction analyzing. The facility topology network model was constructed based on the link and trait point definition.A complete agent navigation model was constructed based on Montello navigation theory, micro pedestrian traffic behavior and finite state machine model, and algorithms of behavior model in different level were proposed. Considering the limited vision of pedestrian agent,the velocity of current path as the main factor to route choice behavior, a quasi-static route choice model was proposed in this paper. Cooperation strategy of agents in the normal pedestrian traffic flow was concerned, a dynamic target updating algorithm for agent in the passageway was proposed for resolving the deadlock of pedestrian agent at the curve conduit. By the proposed crowd holonic search method in this paper, dynamic crowd obstacles avoidance algorithm for pedestrian agent who in the little value pedestrian flow direction in the traffic interweave areas was proposed for resolve the deadlocks appearance in the interweave areas.Finally, considering the contiguous space and discrete time data of simulation system output, statistic method of pedestrian traffic flow parameters and density dynamic showing algorithm were proposed. The pedestrian social force model was validated based on field data and output data of PSSITH simulation system by using the Kolmogorove-Smirnove test method. The social force model which calibrated in this paper was compared with the related research data by analyzing the mean standard error of simulation and real data based on netlogo multi-agent simulation platform. The transfer time mean value of pedestrian in passenger transfer hub was analyzed by using student’s test method based on the real data which were collected from field. The chaotic characteristic of simulation short term pedestrian traffic volume was analyzing by using Lyapunov exponent index. Two real case were analyzed based on the PSSITH, bottlenecks analyze and pedestrian traffic demand loading methods were proposed in the case.This paper will enrich the pedestrian traffic behavior theory and improve the methodology, the navigation model and algorithm proposed in this paper can offer the key algorithm and theoretical foundation, so that researchers can analyze complex evolution process of pedestrian traffic flow more specifically.更多还原