服务水平约束下基于可调节鲁棒优化的供应链分销网络设计模型  

Supply chain distribution network design under service level constraint using affinely adjustable robust optimization

作　　者：孙艺萌 邱若臻[1] SUN Yimeng;QIU Ruozhen(School of Business Administration,Northeastern University,Shenyang 110169,China)

机构地区：[1]东北大学工商管理学院,辽宁沈阳110169

出　　处：《管理工程学报》2021年第3期158-171,共14页Journal of Industrial Engineering and Engineering Management

基　　金：国家自然科学基金资助项目(71772035);辽宁省兴辽英才计划项目(XLYC1907104);中央高校基本科研业务费项目(N180614003)

摘　　要：在不确定需求和多周期环境下,研究了由多个分销中心和多个零售商组成的二级供应链分销网络设计问题。针对需求不确定性,采用预算参数不确定集进行建模;针对分销中心对零售商供货问题,采用联合机会约束进行刻画,用于度量分销网络的服务水平要求。在此基础上,给出了联合机会约束下基于可调节鲁棒优化的供应链分销网络设计模型。在线性仿射决策规则下,运用对偶理论将所建模型等价转换为易于求解的数学规划问题。特别地,考虑到预算参数在模型鲁棒性和解的保守性之间的调节作用,给出了能有效降低分销网络运作成本并同时满足预设服务水平的最优预算参数优化方法。最后,通过数值算例验证了所建模型和求解方法的有效性。结果表明,基于文中方法得到的供应链分销网络具有动态性特征,相应运作方案能够有效抑制需求不确定性扰动,并以最低成本满足预设服务水平要求。此外,与已有凸逼近和样本平均近似方法对比发现,文中模型及方法在网络总成本、联合服务水平和模型求解时间等综合方面更具优势。The rapid change of consumers′preference and the fierce competition of market make it difficult for firms to predict future demands or obtain full knowledge of demand distributions,which results in the low efficiency of the supply chain distribution network.As reported by Aberdeen Group in 2007,two thirds of 143 firms surveyed could not effectively manage the supply chain distribution systems,making them suffer from high operational costs and low service levels.Thus,it is significant to design an efficient and flexible supply chain distribution network which can resist the demand uncertainty and provide a high service level with a low operational cost.In this paper,the design of a two-echelon supply chain distribution network that comprises multiple potential distribution centers(DCs)and retailers is studied under uncertain demands and the multi-period setting.A budgeted uncertainty set which can make trade-offs between robustness and conservativeness is constructed to define the demand uncertainty,and a joint chance constraint which ensures that the demand of each retailer in each period must be satisfied simultaneously with a given probability is adopted to describe the service level of the supply chain distribution network.Considering the fact that the decisions of some variables,such as the ordering quantity and the transportation quantity,can be made after some demands have been observed under the multi-period setting,the affinely adjustable robust optimization is utilized to make sure that the adjustable variables are decided based on the past realized demands.After that,a supply chain distribution network design model with a joint chance constraint is developed based on the affinely adjustable robust optimization approach.The objective of the proposed model is to minimize the total robust cost of the distribution network,including the fixed cost of setting up DCs,the fixed cost of setting up links between DCs and retailers,the ordering cost and the inventory holding cost of DCs and the transportation cos

关 键 词：供应链分销网络设计 需求不确定性 服务水平 可调节鲁棒优化 联合机会约束 

分 类 号：F253.4[经济管理—国民经济]