Abstract: The problem of designing a multi-product and multi-period supply chain network consisting of suppliers,plants,distribution centers and final customers was studied under the uncertainties in both cost and customer demand.For cost and demand uncertainties,the scenario tree method was used to model the uncertainty.A robust design model of supply chain network was established under both the solution robustness and model robustness by taking the supply chain operation performance as the objective function,the supply chain network structure design,logistics operations and so on as constraints,the network nodes selection,linkage path,production,transportation and inventory as decision variables.In view of the large scale caused by large number of uncertain scenarios,the scenario reduction technique was used to filtrate the scenarios,which reduced the solving difficulty.Some numerical examples were executed to validate the stability and effectiveness of the proposed robust supply chain network design model.The results showed that the proposed supply chain network design model could effectively deal with the uncertainties in cost and demand while ensuring the whole network performance,which in addition offered more choices for the decision maker with different risk preferences.

Key words: supply chain network design, robust optimization, uncertainty, risk, scenario reduction

摘要： 在成本和客户需求不确定的条件下,研究了由供应商、工厂、分销中心和最终客户构成的多产品、多周期供应链网络设计问题。针对成本和需求不确定性,采用情景树方法进行建模。在同时考虑解鲁棒和模型鲁棒基础上,建立了以供应链运作绩效最大化为目标,网络结构设计、物流运作等为约束条件,节点选择、连接路径等网络要素和生产、运输、库存等物流量为决策变量的鲁棒供应链网络设计模型。针对由不确定情景数众多导致的模型规模过大的问题,通过情景缩减技术进行了筛选,降低了求解难度。最后,通过数值算例验证了文中所建鲁棒供应链网络设计模型的稳定性和有效性。结果表明,所提出的供应链网络设计模型在确保总体绩效的同时,能够有效应对成本和需求的不确定性扰动,并为具有不同风险偏好的决策者提供了更多可供选择的方案。

关键词: 供应链网络设计, 鲁棒优化, 不确定性, 风险, 情景缩减