【作者】 胡清梅；

【导师】 方卫宁；

【作者基本信息】 北京交通大学 ， 载运工具运用工程， 2011， 博士

【摘要】 摘要：轨道交通车站的设计规划和运营管理是交通工程、建筑规划以及公共安全管理等相关应用领域的重要研究课题。目前在轨道交通车站遭遇大客流冲击时,由于设计者和运营管理者未能有效地根据客流特性来评估轨道交通车站的客流承载能力,存在由车站设计不合理、运营管理措施不到位而引起的客流滞留、客流拥堵、客流混乱乃至踩踏等问题。围绕这些问题,本文以乘客运动特性为出发点,轨道交通车站内的行人设施服务水平为判断依据,提出基于系统动力学建模和客流模拟相结合的轨道交通车站客流承载能力动静态评估方法,预期成果可丰富城市轨道交通车站服务设施的评价和优化管理的研究思路,为轨道交通车站设计者和运营管理者提供理论数据支持。首先,本文选取典型的轨道交通车站,进行大量实地观测和录像,并开发了半自动行人交通数据视频分析系统,系统分析了轨道交通车站的乘客交通特性,包括乘客基本属性、客流到达特性和乘客交通流特性。针对不同类型的轨道交通车站,提出了乘客到达时间集中型的乘客进出站模型、换乘模型和乘客到达时间均匀型的乘客进出站模型,并运用G-S指数和Gini集中度指标评估车站内客流的集中程度。建立了乘客交通流“速度-密度-流量”关系模型,揭示客流运动的基本规律。根据北京市轨道交通车站的乘客交通流特性,以中国行人拥挤感受阈值为划分标准,制定了兼顾交通流特性和乘客主体的服务水平等级划分方法。以调研分析得出的乘客交通流特性规律为基础,结合SP调查结论,提出了适合我国乘客特性的轨道交通设施通行服务水平的划分标准。该标准可为国内城市轨道交通乘客交通设施设计和评价提供重要的借鉴和指导。鉴于轨道交通车站环境的复杂性和乘客运动的多样性,本文提出了一种基于认知的轨道交通乘客运动仿真模型。该模型从人的认知决策行为出发,借鉴Agent建模思想,对乘客行为进行系统的抽象,建立乘客视觉感知模型、乘客行为控制模型、乘客运动路径模型和乘客移动模型,并构建面向智能行为的轨道交通环境知识模型,以实现轨道交通车站的客流仿真。最后运用宏观现象分析法验证了模型的合理性和有效性。该模型集人工智能、元胞自动机、社会力模型和最优化算法的优势于一体,能形成更复杂的乘客运动行为的过程,得出一般的解析模型无法得到的乘客群体运动混沌现象。论文还定义了轨道交通车站客流承载能力的概念,以乘客流特性研究数据和设施通行服务水平标准为依据,从静态计算和动态模拟两个方面出发,建立了轨道交通车站客流承载能力评估与仿真方法。提出了基于轨道交通车站设计规范的车站理论承载能力计算方法、基于系统动力学的车站安全承载能力评估方法和车站客流模拟评估方法。该评估方法一方面可以计算车站安全承载能力,同时还可以模拟分析各类输入变量和管理措施变化时,车站内部的客流密度变化情况,实现轨道交通车站客流的连续动态仿真分析,为轨道交通车站的客流安全管理提供技术支持。最后,本文运用所建立的车站客流承载能力评估理论,计算评估了北京站和复兴门站的客流承载能力,研究了大客流情况下轨道交通车站内瓶颈传播的过程。论文通过案例检验了模型在应用中的能力,成功地再现了车站内部客流的运动特征和车站的空间利用情况,解决了实际中车站客流的运营管理问题,表明所建立的模型具有良好的仿真能力和应用前景。更多还原

【Abstract】 The design, planning and operational management of rail transit station is an important research topic of traffic engineering, construction planning, public safety management and other related fields. Under the circumstances that rail transit stations encounter large amount of passengers, as designers and managers cannot effectively assess the passenger-carrying capacity of rail transit stations according to the passenger traffic characteristics, some problems such as stranded passengers, traffic congestion, traffic chaos and even stampede may occur in that the design and operation management of station is unreasonable. Focus on these issues, using the motion characteristics of passengers as a starting point and the level of service (LOS) of pedestrian facilities as judgment standards, the thesis presents a combination method of static and dynamic evaluation of passenger-carrying capacity based on system dynamic modeling and passenger simulation. Expected results can enrich research ideas of service facilities evaluation and optimal management of stations, and also provide supporting theoretical data for station designers and managers.Firstly, several on-site observation and video-recordings are done in some typical rail transit stations. The thesis develops a semi-automated video analysis system for collecting pedestrian traffic data and analyzes the traffic characteristics of passengers systematically, including the basic attributes of passengers, arrival features and traffic features of passenger flow. For different types of rail transit stations, the ingress and egress models for passenger flow are built for forecasting on pedestrian flow distribution. Meanwhile, G-S index and Gini centralization index are used to assess the centralization degree of passenger flow in stations. The "speed-density-flow" model of passenger flow is built to reveal the basic rule of passenger movement. It not only provides the basis for the planning, design, and operational management of stations, but also provides basic supporting data for the subsequent LOS evaluation of station facilities and passenger flow simulation.According to passenger traffic flow characteristics and passenger crowding feelings, a LOS classification method considering both traffic flow characteristics and passengers is worked out. Based on SP survey results, the proper LOS indicators of station facilities which suites characteristics of Chinese passenger is presented. The indicators can provide important guidance for the design and evaluation of urban rail transit station facilities.Considering the complexity of rail transit station environment and diversity of passenger movement, the thesis presents a cognitive-based simulation model of passenger movement. Using people’s cognitive and decision-making behavior as a start point, based on the Agent modeling theory, it builds passenger visual perception model, passenger behavior control model, passenger movement route model, passenger movement model, and environment knowledge model orientated to intelligential behaviors in order to achieve passenger simulation. At last, macro-phenomena analysis method is applied to verify the rationality and effectivity of the model. The model takes the advantages of artificial intelligence, cellular automaton, social force model and optimization algorithm. It can form the more complex passenger movement process, and obtain the chaotic phenomena that general analytical model cannot get.The thesis also defines passenger-carrying capacity of rail transit stations. According to passenger flow characteristics and the LOS indicators, an assessment method on passenger-carrying capacity of rail transit stations is built based on static calculation and dynamic modeling. The method includes a theoretical calculation method based on rail transit station design standard, an assessment method on safe carrying capacity of stations based on system dynamics and station passenger flow simulation methods. Moreover, it also can achieve continuous dynamic simulation and analyze the changes of passenger flow density in stations while different input variables and management measures change. And then it provides technical support for the safety management of passenger flow in rail transit station.Finally, the assessment theory is applied to compute and evaluate the passenger-carrying capacity of Beijing station and Fuxingmen Station, and also to study the bottleneck spread process in the case of large passenger flow. Through case studies, the thesis reproduces the passenger flow movement phenomenon in stations and station space utilization successfully, and solves the actual operational management problems of the stations. It indicates that the model has good simulation capability and application prospects.更多还原