【作者】 胡晓龙；

【导师】 陈森发；

【作者基本信息】 东南大学 ， 系统工程， 2005， 博士

【摘要】 随着全球经济的一体化,集装箱运输在经济贸易中所体现出来的重要性越来越显著,伴随着集装箱运输的飞速发展,集装箱空箱调运问题也日益突出。关于空箱调运的问题已引起了众多学者的重视,并进行了相应的研究,但就目前的研究水平来看,目前仍处在定性研究阶段,针对空箱调运问题的系统化的、严谨的定量分析目前在国内外文献上鲜有发表。航运公司也大都还沿用经验调度的方式,在没有相应的调箱优化系统支持的情况下,仅仅凭经验要做到科学、合理的调箱非常困难。随着集装箱运输实践的深入和时间的推移,提出更加符合实际集装箱运输网络的空箱调运的理论、模型和方法将是十分必要的,且对于整个ITS的发展有着深远的理论和实践意义。本论文从航运公司的角度出发,借鉴火车的空车调度,将定量分析与定性分析相结合,力图建立科学、合理的空箱调运优化模型,实现空箱调运地点、数量和时间上的最优化。全文的主要研究内容如下:(1)从集装箱的流转过程入手,对空箱调运系统进行了界定,将空箱调运问题分为陆上的空箱调运及海上的空箱调运,并分别从系统的构成、空箱的需求及供应的角度对它们进行分析,建立了以运输费用最小为目标的混合整数规划模型,同时,结合实例对空箱调运的费用进行了详细的分析;(2)分析了实际调运中经常遇到的不确定条件下空箱调运问题。针对陆上空箱调运系统内存在随机性需求的情况建立了机会约束空箱调运模型,探讨了目标函数中含有随机参数的空箱调运模型的求解方法,提出了两阶段随机模拟的遗传算法对调运模型进行求解,通过仿真分析,并与确定性条件下的空箱调运结果进行了对比,表明了模型求解的可行性,符合客观调度实际,为探讨随机陆上空箱调运问题提供了新的思路;(3)将时间窗的特点引入集装箱空箱调运问题中,研究了带时间窗的集装箱空箱调运问题,即当集装箱发送时间有具体限制条件下,如何合理调度空箱的问题。所建立的带有软、硬时间窗的空箱调运的混合整数规划模型符合客观实际,并首次采用免疫算法进行求解,仿真试验表明,该算法能较好解决这一带有复杂约束条件的空箱调运问题,为集装箱空箱调运模型的求解探寻了新的方法;(4)研究了比陆运空箱调运系统更加复杂的海运空箱调运系统,在现有研究成果基础上,对应用Petri网研究海运空箱调运问题进行了的探讨,对长期以来困扰航运公司的空箱调运问题进行了定量分析。建立了基于Petri网的海运集装箱空箱调运模型,提出了遗传算法和Petri网相结合的海运空箱调运求解方法。该方法克服了单一Petri网模型优化处理的局限性和复杂性,充分发挥了Petri网和遗传算法各自的优势和互补作用,使Petri网模型和遗传算法快速求解的优势得到发挥。通过对某航运公司中某航区内基本港间的空箱调运过程的仿真分析表明,应用遗传算法和Petri网对海运空箱调运系统进行仿真求解是可行的,可以快速、合理地确定空箱调运最优的数量、地点以及时间,具有一定的实用参考价值,值得深入研究;(5)充分考虑了客户的不同需求,将集装箱空箱调运问题归结为如何在满足客户需求的基础上通过确定合理的运输方式和运输路径以使总费用最小化。根据调运实际,分析总结了影响客户满意度的因素,并将其归结为6个指标(即及时、快速、准时、方便、经济、适合)。建立了基于客户满意度的空箱调运模型,同时,将灰色评价理论和广义费用的概念相结合,提出依据满意度指标将原费用矩阵转化为广义费用矩阵来确定基于客户满意度的空箱调运方案的方法,该方法能够满足不同客户的特殊要求,自适应地为客户提供最优调运方式,从而,从营销的角度分析了航运企业在愈加激烈的竞争中需要转化经营理念,以增强其竞争力;(6)结合人工智能、Agent模式和集装箱动态跟踪系统的发展,提出了把多Agent模式应用于空箱调运决策系统中,可以在满足在重箱运输需求下,进行多目标的空箱调运优化。提出了基于MAS的空箱调运分层递阶控制系统,该系统具有分层的信息处理和决策机构,可以改善系统各组成部分的协作能力和主动性,提高整个空箱调运系统的决策能力,对集装箱多式联运的发展具有重要的意义。更多还原

【Abstract】 With the globalization of economy, it is more and more obvious that container transportation plays an important role in the world trade. Along with this development,the empty reposition becomes barrier of developing further.This barrier arouses many scholars’interest, and they proceed homologous research. But reviewing the current researching, the study of reposition optimization is still placed in the stage of determining the nature, there are seldom literatures which is dead against systematical and precise quantitative analysis. Shipping company still distributes the empties with the experience. Therefore, it is very difficult to operate economically without mathematics model of reposition.Along with the container transportation development, it is necessary to put forward theory, model and method to match the current transportation net’s development. Furthermore, it is helpful to the development of ITS.From the angle of the shipping company and using the empty car distribution theories for reference, the optimization model which combined the qualitative analysis and the quantitative analysis is constructed for empty reposition , so that the optimization of routing, quantity and time is available. The contents and innovative points of this dissertation include:(1)Through studying the procedure of container delivery, the whole systems of empty reposition is assorted of the land-reposition and the sea-reposition. According to the component of this system, the demand and supply of empties, the homologous mathematics model which can minimize the transportation cost is constructed. Simultaneously, .the cost of reposition is analysed with the real example.(2)Analysing the reposition with unconcerned situation and the random requirement in the land-reposition, a chance constrained model is constructed. On the same time, the strategy of the reposition model which comprises random parameter is found. Within the research, GA of stochastic simulation in two phases to resolve is used. According to the simulation analysis and compare with the data that get from the reposition with certain situation, the model is available. So this model provides new strategy to the empty reposition.(3)The reposition model with time-window is used into resolving the reposition, which can resolve the problem of reasonable reposition with concrete restricting term of launching time. This model with time restriction is accordant with the fact. It is the first using immune algorithm to solve this model. The following simulation shows this algorithm can solve the reposition with complex restriction.(4)Researching the system of sea-reposition which is more complex than land-reposition, how to usePetri net into the reposition is discussed. Therefore, the reposition which troubles shipping company long time can be researched with quantitative analysis. The reposition model which is based on Petri net is constructed and can be resolved with genetic algorithm and Petri net. This method overcomes the limitation and complication of Petri net, and makes the most use of GA and Petri net.The procedure of a certain shipping company’s reposition among its base ports is simulated, which shows that it is feasible to use Petri net and GA to research this problem. This method is useful because it can construct model easily and simply, so that it is worth to research further.(5)Considering the clients’different demand, the problem of reposition is reduced as the cost minimization which is based on the clients’demand through choosing the reasonable routing and transportation mode. The factor which can infect the clients’satisfaction is analysed and introduced as 6 indexes. The empties reposition model which is based on the customer satisfaction through using gray system theory is constructed. On the same time, this model combines the gray system and generalized cost matrix and provides a new way which depend on the indexes. This way can satisfy with different clients and different requirement and provide the reposition optimization.(6)With the development of artificial intelligence, Agent mode and container tracking system, it is the first time to use multi-Agent mode in the decision-making system of reposition. Since the reposition is the object with the logical and physical discrete character, the automatic adaptive cooperative multi-agent tech is used into reposition system to match the laden container’s transportation. Since the multi- agent system has layered information handling and decision-making institution, it can improve the cooperation ability and autonomy of each part, enhance the decision-making ability of whole reposition which is important to the development of container multi-model transportation.更多还原