Abstract: Aiming at the multi-echelon serial supply chain under stochastic multiple demand classes and order splitting,an optimization model for serial supply chains control strategy based on (R,T) inventory policy was proposed.For the complexity of solution,an integrated solution method based on Mesh Adaptive Direct Search (MADS) and simulation was developed.In this method,the serial supply chain system constructed by Arena was responsible for conducting the discrete-event simulation and evaluating the performance of system;MADS implemented by MATLAB was employed to solve the control strategy optimization model.Experimental result showed that the proposed integrated solution method could effectively optimize the control strategy of serial production system and reduce the average total cost by comparing with the genetic algorithm,and its validity was also verified.

Key words: multi-echelon serial supply chain, order splitting, (R,T) inventory policy, mesh adaptive direct search, simulation

摘要： 针对存在多类随机需求及允许订单拆分的多级串行供应链系统,对包括生产策略和配给策略的控制策略进行研究,建立了基于(R,T)库存策略的控制策略优化模型。鉴于求解的复杂性,提出一种基于网格自适应直接搜索与仿真的集成求解方法。集成求解方法中,利用Arena构建的串行供应链系统进行离散事件仿真并返回系统绩效指标;使用MATLAB实现的网络自适应直接搜索,对控制策略优化模型进行优化求解。实验结果表明,与遗传算法相比,所提集成求解方法能有效优化串行供应链系统的控制策略、降低平均总成本,从而验证了该方法的有效性。

关键词: 多级串行供应链, 订单拆分, (R, T)库存策略, 网格自适应直接搜索, 仿真