网络化制造环境下离散供应链系统预测控制策略

针对离散供应链系统中客户需求的高度不确定性，提出最小方差控制方法。该方法相对于传统的预测控制策略，更适合于追踪需求变化减少或消除“牛鞭效应”。首先，将具有z变换的各单元传递函数整合成一个闭环传递函数，以此为整个供应链网络建模，采用白回归移动平均模型描述客户需求趋势，并通过客户需求预测确定两种库存目标水平。其次，建立基于订单策略的目标函数，利用最小方差预测器处理客户需求，提出了供应链的性能指标函数和“牛鞭效应”的分析方程，并通过设定最小方差控制器参数调节超额库存和未交付订货。最后，仿真结果表明最小方差控制在预测市场变化、追踪客户动态需求和保持合理库存水平上是可行和有效的。

Predictive Control Strategy of Discrete Supply Chain System under Networked Manufacturing

Minimum variance control was proposed for high uncertainty of customer demand in discrete supply chain system. Relative to traditional predictive control strategy, this approach is fit for tracking demand variations, and reducing or eliminating "bullwhip effect". The entire chain could be modeled by combining each unit transfer function with z-transform into a close loop transfer function for the network. Customer demand trends were described by a general auto-regressive integrated moving model, and the customer demand forecast was used to determine two inventory target levels. The objective function based on ordering policy was established, and minimum variance predictor was used to handle the customer demand. The performance index function and the analytical equation for the bullwhip effect were proposed, and two control parameters were used to regulate the excess inventory and backorder in minimum variance controller. The feasibility and validity of this approach which forecasts the changes in the market, track customer demand, and maintain a proper inventory level were shown by the simulation results.