【作者】 左大杰；

【导师】 王慈光；

【作者基本信息】 西南交通大学 ， 交通运输规划与管理， 2010， 博士

【摘要】 预计到2020年,我国铁路将逐步形成由客运专线、城际铁路和既有提速线路相配套的、总里程约5万km的铁路快速客运网络。制定面向整个铁路快速客运网络的旅客列车开行方案,需要确定列车的始发站、终到站、列车类别、列车定员、运行径路、停站序列、开行数量等要素,是一类基于复杂网络的大规模组合优化问题。本文将车站其划分为集结站或非集结站,将网络中的客流按给定规则向集结站集结；设计了带有随机客流OD和柔性约束的旅客列车开行方案机会约束目标规划模型和算法；研究了网络中通过能力限制区段等特殊情况下的运输组织方法；采用初步开发的旅客列车开行方案决策支持系统对一个自构造网络进行了算例分析与方案比较。论文的主要研究工作包括以下几个方面：(1)铁路快速客运网络的抽象描述在深入分析铁路快速客运网络中车站、线路、客流、列车、径路等因素的基础上,考虑车站在网络中的地位、作用等因素,将车站划分为集结站与非集结站,并将网络中的客流按给定规则向集结站集结,从而为旅客列车开行方案的优化创造了有利条件。(2)铁路旅客列车开行方案优化模型兼顾铁路、旅客以及其他各方利益,以极大化铁路企业经济效益、极小化旅客出行成本等为优化目标,以通过能力约束、列车运用约束、服务水平约束等为约束条件,建立了旅客列车开行方案优化的确定性多目标硬约束模型；将目标函数和软约束转换为目标约束,根据旅客列车开行方案多目标多属性群决策问题中目标优先级与权重的确定方法,通过引入优先因子、正负偏差量及权重,建立了旅客列车开行方案优化的确定性目标规划模型；进一步以路网中站间OD流量作为随机参数,在一定的置信水平下,建立了旅客列车开行方案优化的机会约束目标规划模型。(3)铁路旅客列车开行方案优化算法考虑站间客流OD的随机性,对机会约束目标规划模型设计了基于随机模拟的遗传算法,主要包括：基因的自然数编／解码、基于客流OD随机分布的初始种群生成、基于序的评价函数计算、遗传算子设计,并采用了自适应的交叉、变异算子保证种群多样性,避免算法早熟现象的产生。(4)特殊情况下的运输组织方法分析车站能力、区间能力等约束条件未能严格满足情况下的运输组织方法,重点研究通过能力限制区段的运输组织方法。阐述了通过能力限制区段的基本理论,针对提速线路、客运专线或城际客运专线三种线路的特点,设计了成组追踪法、分时段运行法以及重联运行法等多种运输组织方法,解决了求解旅客列车开行方案目标规划模型时出现正偏差给运输组织带来的困难,从而使求出的旅客列车开行方案在运输组织上具有可行性。(5)算例分析与方案比较采用初步开发的铁路快速客运网络旅客列车开行方案决策支持系统,基于一个自构造网络进行了算例分析与方案比较。结果表明,所建立的优化模型对铁路旅客列车开行方案决策问题的描述更加全面,而设计的优化算法能够在较短的时间内生成合理的旅客列车开行方案,验证了本文有关模型和算法的有效性和实用性。旅客列车开行方案优化研究是铁路旅客运输系统优化的应用基础研究。论文紧扣铁路快速客运网络中不同车站与客流的重要程度的差异性,描述了客流在网络中的输送机理,设计了旅客列车开行方案的机会约束目标规划模型与算法,研究了特殊情况下的运输组织方法。论文提出的理论与方法形成了一个相对完整的理论体系,不仅对实现铁路旅客列车开行方案的决策优化具有实际意义,同时对丰富我国铁路旅客运输组织理论也具有重要参考价值。更多还原

【Abstract】 By the year 2020, it is expected that railways in China would have been formed into express passenger transportation network, including passenger-dedicated lines, intercity rail lines and existing speed-rising railway lines up to 50,000km in total mileage. Formulating the passenger train operation scheme oriented to the whole express passgener transportation network needs to determine a series of components, such as the departing station, arrival station, train types, fixed members, running route, station orders, etc., which belongs to a large scale combination and optimization problem based on the complex network.On the basis of station positions and functions, and passenger volume, this study divides the stations into muster and non-muster ones, and aggregates the passenger flow to muster stations according to the given rules; Then the chance constraints objective programming models and algorithms are designed with stochastic passenger flow OD and flexible constraints passenger operation scheme. Moverover, the transportation organization methods are studied under special conditions, and the initially-developed passenger train operation scheme decision support system is adopted to exemplify and compare one self-structured network. The main work of the study includes the following aspects:(1)Abstract description of railway express passenger transportation networkBy analyzing the various factors in the transportation network, and considering the stations’status and functions, muster stations and non-muster stations are divided. The passenger flow in the network is clustered according to the given rules, which establishes the hierarchy express passenger transportation network.(2) Railway passenger train operation scheme optimization modelFirst, by giving consideration to railway, passengers and other benefits, deterministic multiple objective contrainted model of passenger train operation scheme optimization is established with constraining carrying capacity, train operation and service level; based on the results, the objective function and soft constraints are transformed into objective constraints, and deterministic objective programming model is set up by introducing priori factor, positive and negative deviation, and weights; Then by considering OD flow within translation stations in the network as random parameter, chance constraints objective programming model is represented at certain confidence interval.(3) Railway passenger train operation scheme optimization algorithmBy considering the randomness of passenger flow OD, stochastic simulatoin-based genetic algorithm is proposed on chance constrain objective programming model, including natual coding/uncoding of gene, initial population generation of random distribution based on passenger flow OD, evaluation function calculation in the form of order and genetic operators. Self-adpative crossing and mutation operators are utilized to make sure the population diversity and to avoid premature convergence of genetic algorithm. (4) Transport organization methods under special circumstancesThrough analyzing the transportation organization methods under unsatisfied condtions with station and section capacity constraints, the methods on restricted section of carrying capacity are studied with emphasis. By explaining the basic theory on restricted section of carrying capacity, and aiming at the features of speed-rising lines, passenger-dedicated lines and intercity passenger special lines, various methods, such as group tracking method, shifting operating method, reconnection operating method, etc. are designed correspondingly to figure out the trouble for transportation organization when the positive deviation occurs while solving the programming model, so as to make the passenger train operating scheme feasible.(5)Example analysis and scheme comparisonThe decision support system of express passenger transportation network scheme is developed initially, and the example analysis and scheme comparison are conducted with the self-structured network. The results indicat that the optimized model can describe the descion-making problems of passenger train operation scheme more completely, while the designed algorithm can generate the suitable passenger train layout and flow distribution schemes in short time, which verifies the effectiveness and practice of the proposed models and algorithms.The optimization study on passenger train operation scheme of railway express passenger transportation network is one of the basic theories research. Concentrated on the difference among various stations in express passenger transportation network and importance extent of passenger flow, this study utilizes various research approaches from qualitative to quantitative, from the local to the global and multi level solutions in order to optimize the whole network and the train flow distribution schemes. The models and algorithms presented in this study reveal the rules of passenger train layout and flow distribution optimization and form into a relatively complete system. The results can not only help to realize the computer-aided decision making of railway passenger train operation scheme, but also enrich railway passenger transportation organization theory in China.更多还原