【作者】 孙彬；

【导师】 陈秋双；

【作者基本信息】 南开大学 ， 运筹学与控制论， 2013， 博士

【摘要】 随着制造业全球化趋势的不断深化,国际集装箱多式联运已经成为国际贸易的重要桥梁,其运输服务网络的覆盖日趋全球化。在这样的背景下,多式联运更加强调物流企业之间的相互协作、以及不同运输方式之间的无缝衔接,以满足不同客户的个性化、一体化服务要求。而如何提高多式联运供应链的运营效率和服务质量不仅引起了人们的广泛关注,而且已成为物流优化领域研究的热点问题。本文旨在从多式联运全局优化、以及局部协调的角度,对上述问题展开讨论。首先在分布式决策下研究企业之间的协商机制、以及多式联运供应链的动态构建方法；然后考虑多式联运中的不确定性、及碳排放等问题,以运输速度为契合点进行扩展研究；最后以集装箱码头的泊位和岸桥为例,讨论了动态不确定环境下物流终端的作业优化对多式联运计划的可靠性以及货物按时交付率的影响。本文具体的研究工作包括如下三个方面：第一、针对多式联运中的动态性、自治性、合作性等特点,提出了一种基于多agent系统的多式联运供应链动态构建方法,旨在通过agent间的信息交互与协商,构建出满足托运人要求的运输方案。构建过程分为招投标决策和中标决策。在招投标决策中,采用重复拍卖机制,通过价格引导使承运人自主压缩运输时间,以满足托运人的时间约束,从而创建出多式联运网络。在中标决策中,通过拉格朗日启发式算法得到满足时间约束的多式联运运输方案。仿真实验表明,针对绝大部分问题实例,这种方法能够有效地求解出满足时间约束下的优化解；同时,相比固定步长下的更新机制,自主决策下的方案更新机制可以有效加快协商过程的收敛性。第二、将运输时间和中转时间的不确定性、以及碳排放等问题引入多式联运供应链的动态构建,建立了多式联运随机模型。一方面,从运输速度和燃油消耗之间的关系入手,采用基于活动的方法估算运输过程中的碳排放量；另一方面,采用机会约束来描述不确定环境下的时间窗约束满足情况,并用近似确定性等价形式代替机会约束,以提高模型的求解速度。在招投标决策中,采用基于动态中转备选集的网络变形、以及考虑期望值的节点预处理来改进上一章的动态协商机制。在中标决策中,针对模型中的机会约束,设计了一种基于变长染色体的改进遗传算法求解模型。通过仿真结果可以看出,针对大部分问题实例,本章推导出的近似确定性等价形式可以有效加快算法的求解速度。仿真实验还分析了机会约束中置信水平的灵敏度,并验证了该模型下的求解方案能够有效地降低运输过程中的碳排放量。第三、针对船舶实际到港时刻的动态不确定性,提出了一种基于鲁棒反应式策略的泊位和岸桥联合调度方法,在分布式决策下设计了多agent系统的框架模型。预调度阶段,在以计划延误时间为服务性指标的泊位分配模型中,引入缓冲时间长度作为鲁棒性指标,建立了基于冗余策略的鲁棒泊位分配模型,并采用CPLEX求解。执行阶段,采用as soon as possible的调整策略和基于合同网的协商机制,合理调配泊位和岸桥,进一步提高系统计划执行的稳定性。仿真实验表明,针对实际中的问题规模,CPLEX可以有效地求得最优解；此外,鲁棒泊位分配方案在不确定环境下的抗干扰能力明显高于传统方案,而相对于仅采用泊位实时调度策略,在执行阶段采用泊位和岸桥联合实时调度策略可以更有效地降低不确定性带来的影响,进一步提高系统执行的稳定性。更多还原

【Abstract】 With the acceleration of manufacturing globalization, the international container multi-modal transport has become an important bridge in the international trade, which is supported by its increasingly globalized transportation service network. In order to provide personalized and integrated service, the multi-modal transport pays more emphasis on cooperation of logistics enterprises and seamless connection of transport modes. The problem of improving the operational efficiency and service quality for supply chain in multi-modal transport has attracted great attention, and become a hotspot in the field of logistics optimization.This dissertation aims to study the above problem through the coordination and collaboration in enterprises, as well as the optimization and scheduling in the logistics terminal. The dissertation includes three research points:(a) the negotiation mechanism and dynamic formation of supply chain in multi-modal transport;(b) decision of transportation speed considering uncertainty and carbon emission; and (c) the impact of resource optimization in logistics terminal on the reliability and on-time delivery of multi-modal transport plan. More specifically, this study can be summarized as follows:Firstly, considering the dynamics, autonomy, and cooperation, a dynamic formation strategy of supply chain for multi-modal transport based on multi-agent system is proposed. The transportation plan is made to meet the demand of consignor through information exchange and negotiation. The formation is divided into bidding decision and winning bidding decision. In bidding decision, the repeated auction mechanism is adopted to create the network of multi-modal transport. To meet the temporal constraint, a price induction mechanism is designed to make carriers reducing their transport time. In winning bidding decision, a Lagrangian-based heuristic algorithm is proposed to obtain the multi-modal transport plan. The simulation experiments show that the formation strategy can provide an effective solution for most of the instances; meanwhile, compared with the mechanism of fixed step size, the autonomous decision can efficiently accelerate the convergence of the negotiation process. Secondly, to deal with the time uncertainty of transport and transit, as well as carbon emission of multi-modal transport, a stochastic model is proposed. Considering the relationship between transport speed and fuel consumption, the activity-based approach is adopted to estimate the carbon emission. The chance constraint is used to describe the satisfaction of time window constraint under uncertainty, and its approximate deterministic equivalent form is obtained to improve the solving speed. To improve the dynamic negotiation of bidding decision in the previous chapter, the network deformation based on alternative set of dynamic transit and node preprocessing with expectation consideration is designed. Considering the nonlinearity of objective function and chance constraint, an improved genetic algorithm based on variable-length chromosome is designed for winning bidding decision. The simulation experiments show that the approximate deterministic equivalent form can efficiently accelerate the convergent speed. Besides, the sensitivity analysis on the confidence level of chance constraint is also conducted and the solution of this model can effectively reduce carbon emission.Lastly, considering the dynamics and uncertainty of vessel arrival time, an integrated scheduling for berth and quay cranes based on robust and reactive policy is proposed. And a model based on multi-agent system under distributed decision is designed. In pre-scheduling stage, using total delay time of vessels as service measure and length of buffer time as robustness measure, the problem is formulated as a robust berth allocation model based on a redundancy policy. And CPLEX is adopted to solve the model. In scheduling stage, for improving the stability of system plan, the as soon as possible strategy and the negotiation mechanism based on contract network protocol are adopted to make full use of berth and quay cranes. The simulation experiments show that CPLEX can solve the optimal solution, and the robust berth allocation plan with real-time scheduling of berth and quay cranes can obviously diminish the effects from uncertainty and improve the stability of system performance.更多还原