【作者】 李静；

【导师】 帅斌；

【作者基本信息】 西南交通大学 ， 交通运输规划与管理， 2020， 博士

【摘要】 随着工业化发展和人们消费方式的转变,货物运输需求结构特征不断发生变化,同时不同运输方式之间的竞争也由成本竞争转向综合服务质量竞争,对铁路运输在时效性,综合服务性等方面提出了更高的要求。由于我国铁路长期以来采用组织型行车组织体制,大部分货运列车需要在编组站进行中转技术作业,车列需要等待凑齐满轴编成辆数才能出发,因此,货车在编组站的停留时间很难确定,也难以保证货物的运到期限,这种困境对编组站的作业模式提出了新的要求。市场竞争条件下的列车编成辆数优化问题研究的是在编组时以运输效益为编组目标,不固定编组辆数,在不超过最大编成辆数(同满轴编成辆数一致)的前提下,通过对列车开行经济成本及效益进行度量寻求最优最小编成辆数,车辆集结到最优最小编成辆数就组织列车编发。该模式打破了列车编成辆数满轴的刚性条件要求,可以提高列车开行的准时性,改善货物运输的时效性,从而为我国铁路货物运输适应现代物流发展、提高货物运输服务质量提供保障。基于此,本文首先重点研究了编组站车辆集结排队系统,并分析了系统性能指标变化,在此基础上,建立了市场竞争条件下列车编成辆数优化模型。论文的主要内容包括以下几个部分:1.首先梳理了列车编成辆数的定义;然后从能力因素、列车运行图、经济因素、运输需求、多运输方式竞争等方面分析了影响列车编成辆数的因素;介绍了刻画车辆在站集结过程的离散时间排队理论并对车辆集结排队系统进行了描述,进而为后续研究提供了理论支撑,并对问题抽象过程进行了刻画与范围界定。2.货车集结是有调车不可避免且耗时较长的技术作业环节,编成辆数与车辆集结延误时间密切相关。针对丢线情况下的放宽条件定点集结模式车辆集结过程,建立了批到达批服务离散时间排队模型。利用嵌入式马尔可夫链方法求得发车时刻车辆集结队长分布,运用补充变量法推得任意时刻车辆集结队长分布与发车时刻车辆集结队长分布的关系式,在此基础上得到车辆平均集结队长、集结延误时间、效率、一昼夜发送车流量等系统指标的数学表达式,通过算例分别分析最小编成辆数、车组大小分布、车流到达强度、服务时间分布对系统指标的影响规律。最后分析了最优最小编成辆数随车流到达强度的变化规律。3.通过对上述研究进行扩展,研究了丢线情况下考虑相邻出发列车车流分配的车辆集结排队系统。首先建立了描述车辆在调车场集结过程的批到达批服务离散时间排队模型,利用嵌入式马尔可夫链方法求得发车时刻车辆集结队长分布,运用补充变量法推得任意时刻车辆集结队长分布,在此基础上得到车辆平均集结队长、集结延误时间、效率、利用率、一昼夜发送车流量等系统指标的数学表达式,通过算例,重点分析了较不分配车流方案分配车流对系统指标的影响规律。4.针对加线情况的放宽条件定点集结模式车辆集结过程,建立了批到达批服务离散时间排队模型,利用嵌入式马尔可夫链方法求得发车时刻车辆集结队长分布,并求得任意时刻车辆集结队长分布与发车时刻车辆集结队长分布之间的关系式,在此基础上得到车辆平均集结队长、集结延误时间、效率、一昼夜发送车流量等系统指标的数学表达式,通过算例分析最小编成辆数、车组大小分布、车流到达强度、服务时间分布对系统指标的影响规律及最优最小编成辆数随车流到达强度的变化趋势。最后与丢线情况的车辆集结排队系统进行了对比分析。5.针对加线情况的放宽条件定点集结模式车辆集结过程,考虑同一去向相邻出发列车车流分配,先编列车在满足集结结束条件基础上,将部分车流编组到后编列车,保证两列列车正点出发,引入车流分配机制。针对该相邻出发列车车流分配问题,建立了描述车辆在调车场集结过程的批到达批服务排队模型,利用嵌入式马尔可夫链方法求得发车时刻车辆集结队长分布,并求得任意时刻集结车辆队长分布,在此基础上得到车辆平均集结队长、集结延误时间、效率、利用率、一昼夜发送车流量等系统指标的数学表达式,通过算例,重点分析较不分配车流方案分配车流对系统指标的影响规律。6.在上述研究基础上,考虑铁路与其他运输方式之间的竞争关系,且不同最小列车编成辆数方案不仅影响铁路广义运输费用函数,还会影响总的运输需求,通过考虑客户对运输方式选择的主要影响因素时间、运价等的基础上,将列车最小编成辆数作为决策变量,构造了一个双层规划模型来描述市场竞争条件下铁路货运市场优化问题。其中,下层规划为多方式弹性需求模型,该模型能够求解均衡状态下的铁路需求量(货流量);上层规划为铁路效益最大化。采用启发式算法求解双层规划模型,下层规划则采用超量需求法进行求解。更多还原

【Abstract】 With the development of industrialization and transformation of consumption structure,the transportation demand structure has been evolving.Meanwhile,the competition among multiple transportation modes has been shifting from the cost competitive stage into the comprehensive service quality competitive stage.They all claim higher requirements regarding the timeliness and comprehensive service quality for the railway transportation.For a long time,the railway in our country adopts the train operation organization system.Most trains require transfer operations in a marshalling station,where cars assemble into a train and depart until there are enough cars available for the departure size limitation.Therefore,the delay time is difficult to predict and the goods delivery time cannot be guaranteed.This predicament has raised new demand for the operation mode in marshalling station.In consideration of the market competition,the optimization of the freight train length aims at seeking the optimal minimum number of cars by measuring train operating costs and benefits.When the number of assembled cars reaches the minimum constraint,the train could depart in the premise that it should not exceed the maximum constraint(i.e.the departure size limitation).It is a break from the rigidity constraint that the assembled car number must reach the maximum before the train departs.It could improve punctuality of scheduled railway service and timeliness of railway goods transportation,and also provide guarantees for the railway transportation adapting to development of modern logistic and improving comprehensive service quality.Therefore,this study aims at investigating the car assembly queuing system and evaluating the system performance firstly.Then,the optimal train length model has also been addressed to maximize the profit of the railway operators in multi transportation modes environment.This dissertation is structured as follows:This paper first teased out the definitions of train length,then provided the influence factors from the perspective of capacity,train diagram structure,economy,traffic demands,competition among multi transportation modes environment.The discrete time queue theory was also introduced,and a car assembly queuing system was described,which was fundament to investigate car assembly queuing models.Finally,research problem and scope of this study were also defined.The car accumulation process is an inevitable and time-consuming railway technical operation,and the train length is closely related to this delay time.In Chapter 3,this study investigated the case that when there are not enough cars available at a departure epoch,the system lost the train path.A batch-arrival and batch-service queuing model for arrival-assembly-departure of cars was built using the method of discrete time queuing theory.Queue length distributions at departure epochs were obtained based on the embedded Markov chain technique.Next,relationships between queue length distributions at arbitrary and departure epochs were derived using the method of supplemental variable.Then,mathematical expressions of various performance of interest were obtained,including the mean queue length,the mean accumulation delay of a car,efficiency,traffic volume in one day etc...And then though numerical examples,the impacts of minimum number of cars,batch size distributions,arriving intensity and service time distributions on the system performance were also analyzed.Finally,how the optimum minimum number of cars changes with the increasing arriving intensity were investigated.Chapter 4 extended the model proposed in Chapter 3.The extensions included the car assignment between adjacent departure trains into the model.A batch-arrival and batch-service queuing model was built for arrival-assembly-departure of cars.Queue length distributions at departure epochs were obtained based on the embedded Markov chain technique.Next,queue length distributions at arbitrary epochs were derived using the method of supplemental variable.Based on these distributions,mathematical expressions of various performance of interest were obtained,such as the mean queue length,the mean accumulation delay,efficiency,capacity utilization,traffic volume in one day.The impacts of car assignment on the system performance were investigated through some numerical examples by comparing the car assignment scenario with the no car assignment scenario.In Chapter 5,this study investigated the case that when at a departure epoch,there are not enough cars available,the system lost the scheduled train path but added a new one meanwhile.For the problem that cars assemble in a marshalling station in the time-fixed mode with soft terms,a discrete time queuing model with batch-arrival and batch-service was established.Queue length distributions at departure epochs were obtained based on the embedded Markov chain technique.Next,queue length distributions at arbitrary epochs were derived.Based on these distributions,mathematical expressions of various performance of interest were obtained,such as the mean queue length,the mean accumulation delay of a car,efficiency,and traffic volume in one day.And then the the impacts of minimum number of cars,batch size distributions,arriving intensity,service time distributions on the system were also analyzed though numerical examples.Changes of the optimum minimum number of cars with the increasing arriving intensity were also analyzed.Finally,comparisons with the model in Chapter 3 were also presented.Chapter 6 extended the model proposed in Chapter 5.The extensions included the car assignment between adjacent trains into the model.In the time-fixed assembly mode with softened terms,the earlier train leaves a part of cars to the later one in the premise that it satisfies the assembly end conditions,so that both trains could depart on schedule.To achieve this,a batch-arrival and batch-service queuing model for arrival-assembly-departure of cars was built.Queue length distributions at departure epochs were obtained based on the embedded Markov chain technique.Next,queue length distributions at arbitrary epochs were derived.Based on these distributions,mathematical expressions of various performance of interest were obtained,such as the mean queue length,the mean accumulation delay of a car,efficiency,capacity utilization,and traffic volume in one day.The impacts of car assignment on the system performance were investigated through some numerical examples by comparing the car assignment case with the no car assignment one.Chapter 7 considered the competition among multiple transportation modes.The generalized cost function and freight volumes were influenced by minimum number of cars.A bi-level programming model was proposed with the minimum number of cars as the decision variable.The lower level program presented a multi-mode equilibrium model with elastic demand,which could offer a railway freight volume.While the upper level aimed to maximize the benefit of railway operator.An iterative heuristic algorithm was developed to solve this model,while the lower level program was solved by excess method.更多还原