【作者】 李娜；

【导师】 李波；

【作者基本信息】 天津大学 ， 管理科学与工程， 2013， 博士

【摘要】 供应链中的生产配送整合问题是目前国内外学术研究的热点，其目标是集成生产排程与配送路径问题，最小化供应链总成本。本论文的工作就是定位于供应链中若干生产配送问题的研究，从供应链全局出发，在供应链运作的各层次上探索生产调度、配送中心定位、配送路径选择及其它们之间关系等诸多问题，包括各种模型建立和相应求解算法的讨论，以及仿真实验验证、比较和分析的过程。研究的工作包括：首先提出了多配送中心-多客户的定位-路径问题，针对配送中心的选择与车辆路径的集成优化，考虑客户需求的模糊性，以配送中心的固定成本和车辆路径成本最小为目标，建立了一个考虑模糊需求的定位—路径问题优化模型。根据求解的需要建立了模糊机会约束规划模型，并转化为具有可变参数的等价模型。采用分解的思想分别给出了定位设施点的成本模型和配送路径的成本估计模型，并分别应用遗传算法和禁忌搜索算法来求解该定位—路径问题。采用的集成优化方法克服了单纯从系统局部角度出发进行研究的缺陷，能有效地解决供应链实践中具有较大规模的生产配送问题。其次建构了一个多周期多层次供应链生产配送批量的混合整数规划模型，主要目的是解决生产订单分配与订购来源决策。考虑生产能力、仓储容量、运输能力等约束条件下，最大化单期订购量。本文采用合并多期需求订购策略，而非过去大部分文献采用的批对批订购策略。提出一个以遗传算法为基础的两阶段启发式求解方法，实验结果表明本文采用的启发式解法与传统求解方法比较，能够得到很好的求解结果良好的求解绩效，也发现合并多期需求订购策略优于批对批订购策略。进一步，针对易腐产品的生产调度和车辆路径问题进行了研究。本文假设零售商的需求是随机的，而易腐产品一旦被生产出来就可能开始逐步变质。优化的目标是最小化供应商的预期总成本。这种模式的目标是最小化供应商的预期总成本，并由此确定最佳的生产数量、开始生产时间以及车辆的配送路线。构建了两种情况下的生产配送模型。一是不考虑车辆启动成本情形下，考虑车辆迟到惩罚成本和产品腐坏成本，应用单亲遗传算法对产品调度编码进行求解，采用传统的插点法解决零售商的车辆路径问题；二是考虑车辆启动成本情形下，同时考虑车辆提前和迟到的时间窗惩罚成本，应用Nelder-Mead单纯形法进行生产调度的求解，采用改进的蚁群算法解决带时间窗的车辆路径问题，结果表明所用的模型和方法能够有效地解决同一框架下的生产调度和车辆路径相结合问题。最后提出了模糊多目标环境下供应链上多成员之间的生产—配送规划决策问题。探讨在模糊环境下，考虑成本、时间和缺货率等限制因素的多目标生产—配送规划问题，建立与求解了含有模糊系数的模糊多目标规划模型，算例验证了模型和方法的可行性与有效性。其次使用了交互式双层规划来求解供应链中一类目标有冲突的模糊多目标规划问题，建立了双层规划的生产配送计划模型，设计了考虑容忍度的模糊交互式求解方法，发展了可调参数两层交互式协调方法，以获得合作满意解。算例仿真验证了该方法的可行性。更多还原

【Abstract】 The integration of production and distribution in supply chain is an importantacademic research hot at home and abroad. Like modern enterprise integratedobjective, the objective of production and distribution is to integrate the productionschedule and distribution route problem and minimize the total cost of productionscheduling and distribution routing problems. So the integration of productiondistribution has become an important task in supply chain management. If theproduction and distribution activities with each other are not coordinated, this willaffect the enterprise operating performance. So the supply chain must developeffective integrated planning to seek solution. This paper focuses on the work ofseveral production distribution problems research. From the angle of the global supplychain and supply chain operation of various levels to explore production scheduling,distribution center allocation, distribution route and their relationship, including allsorts of model and the corresponding solving algorithm is discussed, and thesimulation results verify, comparise and analysis the process.The detail work consistsof:Firstly, coordination of the supply chain in the environment of multi-distributionand multi-customer has been realized, and an optimization model for location-routingproblem is proposed. The integrated optimization of facility selection and vehiclerouting problem is studied by considering the fuzzy demand of customers. In practice,supply chain always operates in uncertainty circumstance, the dissertation establishesthe fuzzy chance-constrained programming model, and transforms it into equivalentmodel with adjustable parameters.And also the genetic algorithm (GA) and tabusearch (TS) are introduced into the location routing problem. Finally, the validity ofthe model and algorithm is demonstrated by a numerical example.Secondly, a novel mixed-integer programming distribution lot-sizing model withtime-varying demand has been developed. And then the paper develops an efficienttwo-phase heuristic method based on the genetic algorithm, in which a combinedmulti-period demand ordering policy, rather than the lot-for-lot ordering policyappeared in the literature is adopted. The experimental results indicate that the goodperformance of the proposed method has been verified through a comparison with theoptimal solution method. It is also shown that the performance of the proposed combined multi-period demand ordering policy is superior to that of the lot-for-lotordering policy.Thirdly, a nonlinear mathematical model to consider production scheduling andvehicle routing with time windows for perishable food products in the sameframework is proposed. The demands at retailers are assumed stochastic andperishable goods will deteriorate once they were produced. Thus the revenue of thesupplier is uncertain and depends on the value and the transaction quantity ofperishable products when they are carried to retailers. The objective of this model is tominimize the expected total cost of the supplier. The optimal production quantities,the time to start producing and the vehicle routes can be determined in the modelsimultaneously. Furthermore, the paper elaborates a solution algorithm composed ofthe constrained Nelder–Mead method and a heuristic for the vehicle routing with timewindows to solve the complex problem. Computational results indicate algorithm iseffective and efficient.Fourthly, this work develops a fuzzy multi-objective linear programming(FMOLP) model with piecewise linear membership function to solve integratedmulti-time period production/distribution planning decisions problems with fuzzyobjectives. The original multi-objective linear programming designed in this workmodel attempts to simultaneously minimize total costs, total delivery time and rate ofshortage. Then an interactive multi-level programming method based on fuzzy theoryfor solving the model is designed. Based on Bi-level programming a new model fornegotiating supply chain cooperation is studied, and fuzzy theory based solutionmethod is also proposed, and an interactive method based on fuzzy theory for solvingthe model is designed in considering the objective tolerance.To solve the problem, itproposes a two-loop interactive algorithm to get satisfactory solution by iterativelyadjusting parameters. In detail, the algorithm includes two interactive procedures: Theformer is major for the preference of the decision maker, realized by fuzzymembership functions reflecting goals and decisions attainments; the latter is for theimprecision of parameters, described by possibility degree. Experimental resultvalidates the feasibility of the algorithm.更多还原