【作者】 狄卫民；

【导师】 胡培；

【作者基本信息】 西南交通大学 ， 管理科学与工程， 2008， 博士

【摘要】 再制造是通过必要的拆卸、检修和零部件更换等,将废旧产品恢复得如同新的一样的过程,它最大限度地重新利用了废旧产品的附加值,降低了环境污染,节约了自然资源,促进了国民经济的可持续发展,正受到许多国家,尤其是工业发达国家的高度重视。适合再制造的产品大多是废旧机电产品,迫于生产商责任延伸制的压力和企业内部经济利益的需求驱动,原始设备制造商(OEM)负责再制造已经逐渐成为一种发展趋势。这使OEM必须面对再制造物流的管理问题。当OEM管理再制造物流时,OEM更便于统筹安排自己的物流活动,基于OEM管理的再制造物流网络也更加复杂,因而值得对其深入研究。本文在已有研究的基础上,针对再制造物流的网络特征,分别研究了OEM管理下的制造/再制造集成物流网络设计的多周期选址模型和不确定环境下的优化设计方法。本文所做的主要工作包括:(1)回顾了国内外有关再制造物流网络优化设计的研究成果,分析了现有研究的不足之处,给出了本文的研究框架。(2)阐述了再制造和再制造物流的内涵,总结了再制造物流网络的特点及其结构特征,对再制造物流网络优化设计的前期准备工作进行了论述。(3)研究了确定性环境下基于OEM管理的再制造物流网络优化设计的多周期选址模型。首先,针对短计划期内不同时间段的设计参数的差异,建立了制造/再制造物流网络优化设计的多周期静态选址模型。该模型综合考虑了传统物流网络和再制造物流网络、及其正向物流设施和逆向物流设施的集成设计问题,利用消费区域的期末新产品或再制造产品的缺货数量、超额供应数量及其未被回收的废旧产品数量实现相邻周期的物流衔接,其运算结果有利于企业战术层次的物流运作管理和相关计划安排。其次,针对长计划期内不同时间段的设计参数的差异,建立了制造/再制造物流网络优化设计的多周期动态选址模型。该模型在多周期静态选址模型的基础上,充分考虑了各项费用的时间价值、设施建设时间的确定以及设施建设时间与设施运营之间的关系,可以实现物流设施的动态选址及其各周期物流量的合理分配,其运算结果有利于企业战略层次的物流运作管理和相关计划安排。最后,考虑到物流设施能力扩展的可能性,在上述多周期动态选址模型的基础上,建立了制造/再制造物流网络优化设计的多周期动态选址扩展模型。该模型综合考虑了设施动态选址、设施动态扩建和设施能力动态扩展问题,不仅能够确定每个周期需要开设的设施位置、数量以及相应物流量的合理分配,而且可以确定每个周期需要扩建的设施位置、数量以及各扩建设施需要扩建的标准系列数量。鉴于上述模型均为混合整数非线性规划模型,本文还研究了求解各模型的混合遗传算法,并进行了算例分析。(4)研究了不确定环境下基于OEM管理的再制造物流网络优化设计方法。首先,考虑到新产品需求量和废旧产品回收量的随机不确定性,在MonteCarlo模拟抽样基础上,建立了样本数量决定求解效率的制造/再制造集成物流网络优化设计的两阶段随机规划模型,探讨了模型求解的混合遗传算法,结合样本均值近似方法阐述了获取理想目标值及其可行解的最优值上下界逼近技术,为得到稳健的制造/再制造物流网络,利用大样本对计算获得的可行网络进行了评价,并明确了基于两阶段随机规划的制造/再制造物流网络优化设计步骤。其次,针对样本均值近似方法的局限性和业主对拟建物流网络的切实要求,引入网络适用概率和网络适用置信水平概念,对既有样本均值近似方法进行了扩展。扩展后的样本均值近似方法充分考虑了给定置信水平下的模型最优值的计算精度,可以得到满足业主要求的随机规划问题的稳健解,具有很好的可操作性。第三,选取产品的绝对数量作为物流分配的决策变量,将消费区域的产品需求量、废旧产品回收量及其各工厂或各中间设施的最大生产或处理能力视为三角型或容差型模糊参数,建立了可以同时进行设施选址与选型的制造/再制造物流网络优化设计的模糊规划模型。该模型的模糊约束置信水平具有很好的独立性,便于模糊机会约束规划方法的应用。为利于模糊约束置信水平的分析和考察,还给出了基于均匀设计的置信水平相对重要性判断方法。最后,基于上述成果阐述了同时含有随机与模糊参数的制造/再制造集成物流网络优化设计方法,从而有效降低了不确定信息带来的决策风险。更多还原

【Abstract】 Remanufacturing is the process of transforming used products into those which satisfy exactly the same quality and other standards as the new products. Remanufacturing is accomplished through necessary disassembly, inspection and repair as well as replacement of components or parts. Because remanufacturing reuses affiliated value of the used product as possible as it can, also plays an important role in reducing environment pollution, saving natural resource and promoting sustainable development of national economy, special attention is given to remanufacturing in many countries, especially advanced industrial countries. Up to now, the product that is suitable for remanufacturing is mostly used mechanical or electronic product. Due to the obligation of extended producer responsibility and the requirement of the corporation’s inner benefit, original equipment manufacturer (OEM) answering for remanufacturing becomes a developing and further tendency, so it is the time that OEM must face and consider the management problems of its remanufacturing logistics. When OEM manages remanufacturing logistics, OEM can plan its all logistics activities as a whole and the logistics network is more complex than traditional network. Therefore, it is worthy to do an in-depth study about this kind of network’s design problems.Based on the existing literatures and aiming at the characteristics of remanufacturing logistics networks, the multi-period location models under certainty and some approaches under uncertainty for optimal design of the manufacturing/remanufacturing (M/R) logistics networks are studied respectively. The main contents of this dissertation are as follows:(1) The developing progress of the optimal design of remanufacturing logistics networks at home and abroad is reviewed, the shortcomings of existing research on the problems are pointed out and research framework of this dissertation is put forward.(2) The connotations of remanufacturing and remanufacturing logistics are discussed, the characteristics and structural features of remanufacturing logistics networks are summarized and the preparative work before the network’s optimal design is expounded.(3) The multi-period location models with OEM management under certainty for optimal design of the remanufacturing logistics networks are studied.Firstly, considering the design parameter’s difference between different periods in a short planning horizon, a multi-period static location model for optimal design of the M/R logistics network is presented. The model solves the integrated design problems by treating traditional logistics network and remanufacturing logistics network as well as forward logistics facilities and reverse logistics facilities simultaneously. It links up the logistics of every inter-related time period into a single stretch by the lack or oversupply amount of the new product or remanufacturing product and the quantity of unreturned used product at every period-end at every consumer zone. The conclusion resulting from the model is suitable for tactical logistics operation management or corresponding schedule arrangement of the corporation.Secondly, considering the design parameter’s difference between different periods in a long planning horizon, a multi-period dynamic location model for optimal design of the M/R logistics network is developed based on the multi-period static location model. This model takes into account time value of expense, establishing time of logistics facility and the relation between establishing time and operation period of logistics facility simultaneously. With the help of the model, appropriate sites of all logistics facility types and sound logistics volume between corresponding facilities of every period can be obtained. The conclusion resulting from this model is suitable for strategic logistics operation management or corresponding schedule arrangement of the corporation.Finally, considering the possibility of logistics facility’s capacity expansion, an extended multi-period dynamic location model for optimal design of the M/R logistics network is proposed based on aforementioned model. The extended model takes into account facility’s dynamic location, dynamic enlarging and facility capacity’s dynamic expanding simultaneously. With the help of the model, appropriate open sites and open numbers of all logistics facility types and associated logistics volume between corresponding facilities of every period can be got, while the sites and numbers of facilities that need expanded and the standard modules that will be expanded at necessary facilities of every period can be achieved. Since these models are mixed integer non-linear programming models, some mixed genetic algorithms are presented and associated numerical examples are given respectively.(4) Some approaches with OEM management under uncertainty for optimal design of the remanufacturing logistics networks are studied.Firstly, considering uncertainty of new product’s demand and used product’s return, a two-stage stochastic programming model is developed by the way of Monte Carlo simulation sampling techniques. The model’s computation efficiency is administrated by sample quantity and the model can be solved by the calculations of mixed genetic algorithm. According to sample average approximation method, optimal value’s upper and lower bounds approaching techniques to acquire good objective value and corresponding feasible solutions are expounded. In order to get a robust M/R logistics network, the feasible networks resulting from time after time necessary optimizations are estimated by using a group of larger quantity samples and logistics network’s optimal design steps based on two-stage stochastic programming are summarized.Secondly, with respect to shortcomings of sample average approximation method and owner’s desire to logistics network, the sample average approximation method is extended by introducing network’s applicable probability and network’s applicable confidence level. The extended method takes into account calculation accuracy of optimal value under given confidence level. With the help of the extended method, the stochastic programming problem’s robust solution that can meet owner’s desire can be got. Therefore, the extended model has good maneuverability.Thirdly, regarding product’s absolute quantity as decision-making variable of logistics distribution, a fuzzy mixed integer linear programming model is presented. The model treats the quantity of recycled product, the demand of new or remanufactured product and the production or processing capability of logistics facility as the type of triangle or tolerance fuzzy parameters, and can optimize facility open site and facility type simultaneously. The model’s confidence level of fuzzy constraint is independent, so it is convenient for the application of fuzzy chance constrained programming method. In order to appraise the confidence level of fuzzy constraint, an analysis method for judging the confidence level’s importance degree is presented.Finally, the optimal design approach with fuzzy and stochastic parameters for integrated M/R logistics network is discussed based on aforementioned research. These means effectively decrease decision-making risk under uncertain information.更多还原