基于需求分布突变检测的两阶段库存管理方法

Two-stage inventory management based on detection of demand distribution change

在需求分布可能会发生突变的两阶段库存管理中,在无法判断需求是否发生突变的时候,决策者面临决策困难.为了解决这一问题,本文基于决策者关于需求突变的经验和更新的需求观测值,提出了无检测库存管理规则、检测库存管理规则和混合库存管理规则三种库存管理方法.比较这三种库存管理方法下的库存系统运作成本发现:检测库存管理规则下的库存系统运作成本不是在任何时候都比无检测库存管理规则情形下低,而是与需求突变发生的概率以及系统的成本结构有关;而结合无检测库存管理规则和检测库存管理规则优势的混合库存管理规则下的库存系统运作成本在任何时候都要比单独的无检测规则库存管理或检测库存管理规则情形下低.本文论证了需求分布突变检测对于需求突变不确定阶段库存管理的价值,并且给出了一种降低需求突变不确定阶段库存成本的策略.

This paper studies a two-stage inventory control problem where demand distributions may change, and it would be a difficult problem when decision makers can＇t get accurate information of the changes. To cope with it, we develop three inventory management rules, namely, the non-detecting rule, the detecting rule and the hybrid rule, which are based on past demand changes and the updated demand data. We explore the advantages of those rules by comparing the operating costs of different inventory systems adopting different rules. Consequently, we prove that the operating cost under the detecting rule is lower than the cost under the non-detecting rule on the prerequisite of the special relationship between the probability of demand changes and the structure of inventory costs. With the consideration of the prerequisite, we construct the hybrid rule by integrating the non-detecting rule and the detecting rule. Then, we find that the hybrid rule would induce lower operating costs than the two other rules. Our results attest that it is worthy to detect the changes of demand distribution when demand changes can＇t be known with certainty and we suggest strategies for minimizing the operating cost.