供应链网络优化

【作者】 刘诚；

【导师】 陈治亚；

【作者基本信息】 中南大学 ， 交通运输规划与管理， 2006， 博士

【摘要】 在信息化、数字化及网络化的今天，供应链管理充满了诱惑力，它是一个管理时代的象征，是新的生产力，供应链管理带给我们的不仅仅是一种新的管理工具，更重要的是有了更新的管理理念；有了提高认识的机遇；有了籍此重新规划、设计和优化业务流程的途径。供应商的评价和选择是供应商管理的一个非常重要的环节。本文提出了带时间窗的供应商的选择问题。研究了有时间约束且有多个供应商可供选择的前提下，如何给出一个满意的供应商选择方案，确定参与的供应商及各自所供应的物资数量使总成本最小。对该问题在允许缺货和不允许缺货的前提条件下分别给出了单一物资需求和多物资需求的数学模型，并设计了相应的算法进行求解，证明了算法的最优性，给出了算法的复杂度。指出了所设计的算法是拟多项式时间算法，具有良好的性能。供应链设计是实现供应链网络优化管理的一个重要手段。由此本文考虑以核心企业为主导的从供应商—转运中心—制造商—配送中心—分销商的供应链，提出了基于转运中心和配送中心选址的供应链设计问题，给出了以供应链建设和运营整体总成本最小化为目标的非线性规划模型。为了与重要供应商和分销商建立良好的合作伙伴关系，在供应链设计中尽量体现优先向重要供应商采购原料和优先向绩效高的分销商供应产品的思想，在此模型的基础上改进得到了以供应链整体成本最小化、供应商加权原料物流最大化、分销商加权产品物流最大化为目标的多目标规划模型。针对多目标规划模型求解的困难性，将供应链整体成本最小化的目标函数转化为约束条件，然后再利用线性加权法将供应商加权原料物流最大化、分销商加权产品物流最大化这两个目标函数合成为一个目标，将多目标规划模型转化为单目标规划模型，从而给问题求解带来了方便。配送是供应链管理活动的关键环节之一，车辆路径的选择是实现优化配送的一个主要内容。基于车辆路径问题的NP-完备性，提出了一个并行遗传算法对带软时间窗的物流配送车辆路径问题进行求解，与其他相关算法进行比较，表明该算法具有良好的性能。进而考虑到物流配送车辆路径问题中要涉及货物的装卸作业，将装卸工调配问题和车辆路径问题相结合提出了含装卸工调配的物流车辆配送路径问题，给出了以总运输费用最小、总装卸工人数最少为目标函数的双目标整数规划问题的数学模型。按目标函数的主次分两个阶段对该问题进行了求解；并将装卸工人数最少转化为装卸费用最小将该模型进行了推广。最后将车辆路径问题和服务水平相结合提出了物流配送模糊车辆路径问题，以降低配送总费用和提高服务水平为目标，给出了相应的数学模型，设计了一个混合遗传算法对其求解。更多还原

【Abstract】 SCM (Supply Chain Management) becomes even more appealing in an era of information, digitization and internet. As both the symbol of a new management era and a new production force, SCM provides us with not only a new management method but also an updated management philosophy. It also opens our eyes to a new perspective and points to us a way to reprogram, design and optimize the procedure of transactions.The evaluation and selection of supplier play a very important role in Vendor Management. The thesis propose the problem of suppliers selection with time windows,and then discussed how to come up with a satisfying proposal on supplier selection and how to minimize the total cost by getting our goods proportionally from different suppliers on condition that time restriction exists and many suppliers are available. On the basis of allowing goods shortage and not allowing goods shortage, the mathematical models and algorithms to single goods demand and multiple goods demand is presented respectively.The algorithm’s feasibility, optimality and complexity are deeply analyzed. The algorithm is simulated polynomial time algorithm which has premium properties.The design of supply chain is a important means to achieve the optimal management of SCM. According to this, by considering the core enterprise as leading sector through SCM from provider to transshipment center, to manufacturer, to distribution center and to distributor, this article proposes the design problems of SCM based on site selection of transshipment center and distribution center. It shows non-linear programming model which aims to minimize the whole cost of SCM construction and whole operation. In order to set up good partnership with primary providers and distributors, in the design of SCM, one should do their best to reflect the thought of a prior purchase of raw materials from primary providers and a prior supplement of products to distributors with high performance. Through the improvement of this model, a multi-objective model including minimizing whole SCM cost, maximizing providers’ weighed raw materials and distributors’ weighed commodities current of materials is given. As for the difficulties of solving multi- objective model, the objective function of minimizing of whole SCM cost is transferred into constraint condition, and then using one dimensional weighed mediation method combines the two objective functions of maximizing providers’ weighed raw materials and distributors’ weighed commodities current of materials. Consequently, multi-objective programming model is transferred into single objective programming model, which simplifies the problem.Distribution is one of the most important link of the SCM.The selection of vehicle routing is a major method to realize optimal distribution.Based on the NP-complete of the vehicle routing problem,a parallel genetic algorithm is proposed to solve the vehicle routing problem with soft time window of logistic distribution.Compared with other algorithms,the parallel genetic algorithm is a efficient method. Furthermore considering the fact that physical distribution vehicle routing involves the process of loading and unloading goods,this paper combined loader scheduling problem with vehicle routing problem ,a physical distribution vehicle routing problem with loader scheduling is put forward.A mathematical model of double-objective integer programming is given involving minimum total transportation cost and minimum total number of loaders.According to the primary and secondary of the objective functions,solving process of the problem is divided into two phases,transform minimum number of loaders into minimum loading and unloading cost,this paper extended the model.At last,combine vehicle routing problem with service level,a fuzzy physical distribution vehicle routing problem with the objective to reduce total cost of distribution and improve service level is proposed,and a relevant mathematic model is given,a hybrid genetic algorithm is desigened to solve this problem.更多还原