Abstract: Due to the structural complexity and multiple objectives of multi-echelon supply chain,the influence of uncertain factors on its overall performance could not be ignored.To estimate the impact of uncertain factors on its overall performance,the arena simulation and Kriging meta-modeling technology were integrated to construct Kriging mean and standard deviation meta-models of multiple performance response indicators based on the idea of robust parameter design.With the help of the constructed Kriging meta-models and satisfaction function method,a robust comprehensive satisfaction optimization design strategy based on Kriging meta-model was given.The influence of uncertain factors on the robust optimization solution was measured by the nonparametric bootstrap sampling method,which was compared with the robust optimization method based on polynomial model.Simulation results show that the proposed method can effectively solve the robust optimization problem of multi-echelon supply chain with multiple performance response under uncertain parameters.The proposed method could ensure the supply chain system to achieve the desired optimal performance by minimizing the impact of uncertain factors on the supply chain performance,which provided a theoretical basis and decision-making reference for the stable operation of multi-echelon supply chain.

Key words: multi-echelon supply chain, Kriging meta-model, arena simulation, robust parameter design, desirability function approach

摘要： 由于多级供应链的结构复杂性和目标多重性,不确定性因素对其整体绩效的影响不可忽视。为了估量不确定性因素对其整体绩效的影响,首先,将arena仿真和Kriging元建模技术相融合,基于稳健参数设计的思想,构建多个绩效响应指标的均值和标准差Kriging元模型;其次,借助已构建的Kriging元模型和满意度函数法,构建基于Kriging元模型的稳健综合满意度优化设计策略;最后,采用非参数bootstrap重复方法刻画不确定因素对稳健优化解的影响,并与基于多项式元模型的稳健优化方法进行比较分析,仿真验证了所提方法的有效性和稳健性。结果表明,所提方法能够有效解决不确定条件下具有多绩效响应指标的多级供应链的稳健优化设计,在保证供应链系统达到期望最优绩效的同时,尽可能地降低不确定性因素对供应链绩效的影响,为多级供应链的稳健运营提供一定的理论依据和决策参考。

关键词: 多级供应链, Kriging元模型, arena仿真, 稳健参数设计, 满意度函数法