【作者】 刘新华；

【导师】 周伟；

【作者基本信息】 长安大学 ， 交通信息工程及控制， 2013， 博士

【摘要】 随着城市化进程的加快，国内主要城市地铁建设快速发展，各大城市地铁运营网络逐步形成，地铁客流网络化特征日趋明显。与过去的单条或少数几条地铁线路的客流特征相比，网络化运营地铁乘客的路径选择行为以及在此基础上形成的网络客流时空分布特征均发生较大变化。按照目前北京地铁票制和票价政策，由于乘客换乘不需刷卡付费，导致线路路径流量和线路换乘流量的不确定性，给城市地铁运营调度管理和客流组织带来了挑战。因此，对地铁网络中乘客的路径选择规律以及客流分配方法进行深入研究，对提高城市地铁运营精细化管理水平具有重要意义。本文的研究主要围绕城市地铁网络客流分配问题而进行。通过对城市地铁乘客出行特性的调查分析，得到影响乘客路径选择的主要因素以及相关规律，对乘客属性与路径选择行为之间的相关性进行了统计分析，基于调查数据确定了乘客分类模式，特别是提出了基于乘客属性的地铁配流模型及算法，并对算法进行了实证研究；同时，考虑了地铁需求的动态属性，给出了基于时刻表的地铁时空扩展网络的概念和构建方法，在此基础上，建立了基于时刻表的地铁客流动态分配模型及算法，并利用北京地铁网络运营客流数据对模型及算法的可行性和有效性进行了验证。具体而言，论文主要就以下方面进行了研究：（1）为了充分掌握乘客的路径选择行为，同时，为了给后续研究提供可靠的数据支持和相关参数估计，针对北京市地铁运营网络，进行了乘客路径选择行为调查，并对调查数据进行了统计分析。结果表明：不同属性的乘客对路径选择存在明显差异，不同属性乘客对时间和换乘等影响因素的关注程度呈现明显差异。（2）基于调查数据，采用卡方检验方法，对乘客属性和路径选择之间的相关性进行了分析，根据分析结果可知，在调查问卷所涉及到的乘客属性中，只有职业、出行目的和个人收入这三个属性和路径选择行为具有较强相关性。同时，对以上三个属性进行了合并分类，每个属性划分为两大类，并对不同类别属性与路径选择进行了相关性分析。最后，对这三个属性进行交叉分类，产生八个乘客类别，进而对样本数较少的类别进行了合并，最终，将全部乘客划分为五个类别。（3）提出了考虑乘客类别的城市地铁网络流量分配方法，根据调查数据，对模型中的参数进行了估计和验证，并对O-D需求进行了类别划分，最后，通过实际网络数据对模型和算法进行了实证分析。得出如下结论：基于乘客分类的地铁配流方法相对于不考虑乘客分类的方法，其计算结果更接近实际；通过对模型中相关参数的灵敏度分析可知，表示乘客对网络熟悉度的参数i对计算结果影响相对较小，而有效路径判定条件中的扩展系数H以及换乘惩罚参数i和i对计算结果影响较大；通过对所有五类乘客所占百分比的变化对配流结果影响的分析可知，由于第二类和第三类乘客的路径选择偏好与其他类别相比有较大差异，因此这两类乘客所占乘客比例的变化对配流结果的影响较大。这些结论证明了根据乘客属性对乘客进行分类，并对不同类别的乘客重新进行模型参数估计并进行客流分配方法的合理性和必要性。（4）总结了地铁网络的时变特征和O-D需求的时变特征，给出了基于物理站点和区间的地铁网络空间结构的定义和描述，结合列车运行时刻表信息，定义了基于时刻表的地铁时空扩展网络以及时空路径的概念，在此基础上，给出了基于时刻表信息构建地铁时空扩展网络的基本方法，并对该网络中的基本元素进行了定义和描述。（5）综合考虑影响乘客出行选择的主要因素，分析和研究地铁乘客路径选择行为的时效性，基于地铁时空扩展网络，考虑了乘车时间、换乘因素以及车内拥挤因素，构建了基于地铁时空扩展网络的路径广义费用模型，在此基础上，提出了基于时刻表的地铁网络客流动态分配模型，并采用MSA算法进行求解，同时，还给出了针对地铁时空扩展网络的改进最短路搜索算法。最后，通过一个简单算例对模型及算法进行了验证和分析。更多还原

【Abstract】 With the sustained construction of urban subway in China, the urban subway network will beenlarged gradually in metropolis.Compared to single or several subway lines, passengers’ behaviorsconsisting of line choice, route choice and transfer choice changed greatly. In accordance with the Beijingsubway tickets and fare policy, there is no dditional charge for transfer, the challenge to the operation andmanagement of such system is how to predict the nonlinear growth of travel demand and simulate thepassengers’ behaviors consisting of line choice, route choice and transfer choice. In such conditions, it is akey problem for its operation and management to predict scientifically passenger flow assignment throughurban railway transport network.This thesis focused on the passenger flow assignment problem in the urban subway network. Throughthe analysis for the travel survey of Beijing subway passengers, this paper considered the main factorsaffecting passenger’s path-choice and analyzed the correlation between the passenger’s properties and thepath-choice behavior. Based on survey data, the classification of all passengers was determined. Then thesocio-economic attributes based choice probability models are given. Subsequently, the correspondingassignment process is proposed for urban subway network. Similarly, the Beijing subway network in2010is used as example to illustrate the applicaiotn of proposed model and algorithm. At the same time, thispaper considered the dynamic properties of the urban subway network and proposed both of concept and itsmodel of schedule-based subway space-time extended network. On the base of these, a schedule-baseddynamic assignment model and algorithm for subway network was presented with its solution algorithm inthis paper. Finally, the feasibility and effectiveness of the model and algorithm were tested by a simpleexample. The contents of this paper are summarized as follows:(1) In practice, the flow assignment partern through urban rail transport network is resulted with theaggregation of all passengers’ cmbined choices. Therefore, understanding and mastering the passenger’schoice psychology is the the foundation of modeling the passenger’s route choice and flow assignmentthrough whole network. In order to fully grasp the characteristics of the passenger route choice behavior,meanwhile, in order to provide reliable data for future research support and related parameter estimation,this paper designed a questionnaire of passengers’ choice behavior and gives fully data statistical analysisof Beijing subway network.(2) Based on the survey data, this paper analyzed the correlation between the passenger’s propertiesand path-choice behavior by using the chi-square test. According to the analysis results, only threeproperties including the occupation, the trip purpose and personal income have strong correlations with thepassenger’s route-choice behavior. Each attribute was divided into two major categories and eight types ofpassengers were found by cross-classification of these three attributes. Finally, the categories with therelatively small sample size were merged and resulted in five categories of all passengers.(3) According to survey data, the passengers with different socio-economic attributes will havedifferent standards and preferences of traffic choices. This paper considers the different choice behaviors between different passengers fully and establishes the generalized travel cost functions for different typesof passengers. Then the socio-economic attributes based choice probability models are given. Subsequently,the corresponding assignment process is proposed for urban subway network. Similarly, the Beijing subwaynetwork in2010is used as example to illustrate the applicaiotn of proposed model and algorithm. Thefollowing conclusions were draw from this research: the calculation results from the socio-economicattributes based assignment method were closer to the actual case with respect to the method withoutpassenger classification; through sensitivity analysis, it is shown that the parameters that desribed thepassenger’s familiarity of the network had relatively small impact on the calculation results, while theexpansion coefficients for effective path as well as the transfer penalty parameters had more impacts on thecalculation results; due to the second class and the third class passengers had different path-choicepreferences compared to the other categories, the changes of these two types of passengers share wouldimpact the flow assignment results. These conclusions expained that the reasonableness and necessity ofthe socio-economic attributes based assignment method.(4) This paper summed up the time-varying characteristics of both the subway network and O-Ddemand and proposed the definition of the spatial structure of the subway network based on the physicalsite and interval. Combined with the train schedule, this paper presented the space-time extended networkas well as the concept of space-time path. On this basis, this paper proposed a method to build thespace-time extended network for urban subway system based on the schedule information and gave thedefinition and description of the basic elements in such network.(5) The major factors that influence passenger flow pattern in urban subway network are taken intoaccount. The timeliness of the impact on passenger travel choice was analyzed. Based on the space-timeextended network for urban subway system, the factors in term of travel time, transfer as well as thecongestion in vehicle were fully considered in order to build a generalized cost function for passenger’stravel path. On the base of this, this paper proposed a schedule-based dynamic assignment model for theurban subway network. The MSA algorithm was used to solve such model. Simultaneously, an improvedshortest path search algorithm was given for the space-time extended network. Finally, a simple numericalexample was used to illustrate validation and effection of the proposed model and algorithm.更多还原