【作者】 陈崇双；

【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， 2012， 博士

【摘要】 车流组织是铁路行车组织工作中的重要内容,一直也是国内外铁路行业专家以及运筹管理领域学者们研究的课题之一。由于铁路运输组织的行业特殊性以及网络大规模尺度,指导运输生产的各种计划和方案相互关联,相互影响,从而使得车流组织优化非常复杂也非常困难,至今仍未圆满地解决。分组列车作为其中的一种车流组织形式,在以美国和加拿大为代表的铁路货运发达国家非常普遍。但由于铁路管理体制的不同以及运输组织模式的差异等其他方面原因,国外的研究成果难以完全应用于我国实际。而在国内,车流组织研究主要集中于单组列车,很少涉及分组列车；实际生产中分组列车的比例也非常低,目前主要在牵引定数递减情形以及集装箱五定班列中有过采用,但并不广泛。因此,分组列车的理论成果和实践经验还非常缺乏。相比单组列车,分组列车组织较为复杂,要求编成站具有较大的改编能力和较多的调车线路等。尽管如此,仍有其有利性存在,如有利于保证列车满轴正点和不违编,远程车流在换挂站部分改编而停留时间较短,减轻沿途技术站的负担等。既然分组和单组列车各有利弊,那么二者相互配合,扬长避短,发挥各自的优势对提高运输效率和效益将很有现实意义。另外,在列车的构成上,单组列车和摘挂列车都可以视为特殊的分组列车。从这个角度来说,分组列车的内涵更大,其理论研究也更具普适性。因此,对分组列车组织的相关问题研究,具有实际应用价值和学术研究价值。本论文在分析铁路车流组织原理、总结国内外编组计划优化方面的理论和方法的基础上,运用系统工程原理以及微观机理融合宏观机制的指导思想,采用定性分析与定量测度相结合,理论推导与仿真实验相融合的方法,比较系统地研究了分组列车组织特征、集结特性、组织条件、编组方案优化等相关问题,主要研究工作包括以下几个方面：(1)既有文献的对比分析。在总结大量相关文献的基础上,从研究对象、建模方法、求解算法三个方面分别介绍国内外编组计划的研究现状,对比并评述国内外编组计划在编制流程、构成内容和建模方法三个方面的特点。(2)分组列车车流组织特征分析。在对车流组织的含义、内容、组织方案以及我国车流组织经验概述的基础上,分析了分组列车的基本特征,包括技术作业特点,编成站、换挂站和终到站的作业流程,货车构成,利弊分析,组织条件分析以及各种组织形式分类,并与单组列车进行了相应地对比。(3)分组列车集结特性研究。以双组形式又尤以其中的固定重量形式分组列车为研究对象,对其集结特性进行定性分析和定量测度。将固定重量形式分组列车的总集结耗费划分为固有和附加两类,根据集结过程的动态特性和车流到达的不确定性,将到达批中的车辆数和间隔时间都视为随机变量,在独立同分布的假定下,描述固定形式分组列车在编成站和换挂站的集结过程。进一步假定到达批中的车辆数服从泊松分布以及间隔时间服从指数分布,应用随机过程知识,理论推导了集结特性的三个表现方面：集结批次、集结占用时间以及集结消耗,采用数值计算方法分析了单参数的灵敏度以及双参数的耦合影响,并给出三者均值的估计公式。(4)分组列车组织条件研究。也以双组形式的固定车组重量分组列车为研究对象,其组织条件主要包括列车编成辆数最佳分配和开行适用条件。对于前者,将其描述为,在车流到达特征已定的条件下,如果车流量递减,通过确定基本组和补轴组重量的最佳组合,使得平均每列车的集结耗费最小或者集结占用时间最小,建立整数优化模型。模型采用分阶段逐步求解思路,首先通过数值计算探究最佳固定重量的影响机制,然后基于挖掘的信息采用回归方法给出经验计算公式。对于后者,基于固定形式分组列车与单组列车集结耗费的比较分析,以前者相对后者的总净节省作为其综合效益的度量,构建了其开行适用性条件,并采用离散事件系统仿真方法进行验证。(5)分组列车编组方案优化参数。研究了分组列车编组方案的优化参数,具体包括计划车流量,固定和非固定形式的集结参数,部分改编相对无改编通过增加时间、相对完全改编的减少时间两个节省时间参数,途中改编次数,以及最大车流组号数。(6)分组列车编组计划优化模型。以途中仅换挂一次的双组列车为研究对象,在同时采用分组和单组列车形式进行车流组织的框架下,根据编组去向在车流和列流之间的过渡和衔接角色,重构车流组织任务,定性分析综合编组方案的内容和特点。基于车流组织任务的分解和我国的实际情况,忽略列车接续子问题,进而以路网编组去向方案、编组去向接续方案以及路网列流方案为0-1决策变量；考虑物理方面的车站调车线数约束和改编能力约束,车流组织制度方面的车流接续归并和车流不拆散约束,以及决策变量之间的逻辑约束；以列车集结和车流改编的总耗费最小为目标函数,建立车流径路已知条件下综合编组方案优化模型OMITFP。针对模型的特点,对其中编组去向接续子问题(CBAP),以路网编组去向方案和编组去向接续方案为0-1决策变量,考虑避圈约束、最大改编次数约束、决策变量之间的逻辑约束以及筛除显然不利方案,构建数学模型并设计了方案树法和回溯算法。更多还原

【Abstract】 As one of the research hotspots for railway experts and scholars in operational research (OR) and management science at home and abroad, wagon flow organization plays an important role in train operation. Owing to the particularity of railway transportation operation and the large-scale network attribute, kinds of plans and schemes conducting transportation production are correlated and interchanged with each other, which makes this problem so extraordinarily difficulty and complicated that it has not been resolved completely so far.Multi-block train, awagon flow organization mode, is popular and widespread in U.S. and Canada representative of the countries with developed railway freight transportation. However, it’s difficult to fully applied abroad achievements to our country because of the difference in management system, transportation organization model, etc. In China, the research on wagon flow organization mainly focuses on one-block train (single group train) and rarely involves multi-block train, as well as the latter is of a very low proportion in practice and just adopted in traction destiny decline situation and "Five Scheduled" container trains at present. Therefore, theoretical achievements and practical experience of multi-block train are still scarce.Comparing with one-block train, the operation of multi-block train is more complex due to requiring the formation station with larger resorting capacity and more reclassification tracks, etc. However, it also has its own advantages, such as guaranteeing full train tonnagepunctuality and complying with train formation plan (TFP), shortening the transfer delay for long-distance wagon partly-resorting in intermediate block-swap station(s), reducing the load of technical stations along the route, etc. Considering the pros and cons of the two modes, it will be practically meaningful to improve efficiency and benefit of transportation by cooperating, maximizing the favorable factors and minimizing unfavorable ones. Moreover, one-block train and pick-up and drop train can be regarded as special multi-block trains in term of train’s constitution. From this point of view, multi-block train has wider extension, and the corresponding theoretical research is more universal. So it is valuable in both practice and academy to research the related problems of multi-block tram.Based on the analysis the principles of wagon flow organization and the summary of optimization theories and methods of TFP at home and abroad, following the guiding ideology of system engineering and micro-mechanism fusing with macro-mechanism, this paper systematically probed into the problems of operation characteristics, accumulation properties, operation conditions, train formation plan optimization, etc, by adopting qualitative and quantitative analysis together with deduction and simulation method. The main research work is as follows: (1) The comparative analysis of existing literature. On the basis of summing up related literature, the research status of TFP at home and abroad was introduced from the aspects of research object, modeling method and solution algorithm and their respective characteristics were contrasted and reviewed in terms of conducting process, composition content and modeling method.(2) Analysis of the operation features of multi-block train. Base on the overview of definition, content, operation scheme and experience of wagon flow organization, in China, some characteristics including technical operating traits, operation process in formation station, intermediate block-swap station(s) and destination, cars’ categorical attributes, advantages and disadvantages, operation conditions and various types of multi-block train mode were analyzed and contrasted with one-block mode.(3) Study on the accumulation properties of multi-block train. Taking double-block train with fixed weight as the object, qualitative analysis and quantitative measure were conducted aiming at the accumulation properties of multi-block train. Dividing the total aggregation cost into the inherent and additional aggregation cost, according to the uncertainty for arrival and the dynamics for aggregation process, the aggregation process of the fixed mode multi-block train at the formation station and intermediate block-swap station(s) were described by considering the size and interval time of the batch of cars as random variables and assuming them to independent and identical distribution. Furthermore, assuming that the size of the batch of cars obeys a Poisson distribution and the interval time follows an exponential one, the performance of aggregation process, i.e., aggregation batch, aggregation occupation time and aggregation cost were deduced applying stochastic processes theory. With the numerical method, estimation formulas of their mean were obtained while analyzing single parameter’s sensitivity and the coupling influence of double-parameter.(4) Study on the operation conditions of multi-block train. The double-block train with fixed weight was chosen as concern object yet whose operation conditions embrace the determination of train constitution and applicable operation conditions. The former was modeled as determinating the optimal proportion between basic-block (i.e. long-distance block) and complementary-block (i.e. short-distance block) to minimize the aggregation occupation time or aggregation cost given the arrival characteristics of the wagon flow. An Integer optimization model was established and solved though two phases. In phase I, the effect mechanism of optimal weights was mined using numerical calculation. In phase Ⅱ, the corresponding empirical regression formulas were respectively provided according to the explored information. Based on the analysis and comparison of aggregation cost between multi-block train with fixed weight and one-block train, applicable operation conditions of the latter was built by adopting the total net savings between the former and the latter as the measure of its comprehensive benefits and tested using discrete event system simulation (DESS) method.(5) The optimization parameters of multi-block train formation plan. This section studied the parameters for multi-block train formation plan optimization, including the traffic flow in schedule length (planning horizon), accumulation parameters of fixed and non-fixed pattern, two time-saving parameters about block swap that one is relative delay corresponding to passing through and the other is relative saving corresponding to resorting, the up-bound of reclassifi cation and maximum possible blocks in every station.(6) The optimization model of multi-block train formation plan. Taking the double-block train which has only one block-swap on its route as the research object in this section, under the framework by providing multi-block train and the one-block train services at the same time, the task of wagon flow organization has been reconstructed and the contents and features of the integrated train formation plan (ITFP) have been qualitatively analyzed according to the transitional linkage of blocks between cars and trains. Considering the decomposition of the above task and the actual situation of China, the block-to-train assignment plan (BTAP) was ignored, and then the block network design plan (BNDP), car-to-block assignment plan (CBAP) and train network design plan (TNDP) were taken as0-1decision variables.In order to minimize the total delay cost generated by aggregation, reclassification and transfer in all itineraries, an optimization model of integrated train formation plan (OMITFP) in the situation of giving car routing was constructed while satisfying the limitation in physical aspects, i.e., resorting capacity and reclassification tracks for every terminals, restriction on operational rules and regulations for wagon-flow, i.e., non-separation and consolidation (every railcar originating or re-blocked at a yard and destined for another must travel in a block to the same next re-blocking yard), as well as logical constraints among the decision variables, Considering the model’s characteristics, when two0-1decision variables corresponding to block plan and car-to-block assignment were introduced, a model to generating all possible CBAPs was developed with the constraints of non-circuit, maximum frequency of reclassification, cross relation and cutting off unfavorable plans. Besides, the method of scheme tree and backtracking algorithm was designed.更多还原