| Summary: | 碩士 === 國立中興大學 === 企業管理學系所 === 102 === This paper discusses the selection of three kinds of reorder level from the perspective of a cost seeking corporate and a cumulative grade minimizing. Specifically, we consider a Markov Decision Process model for selecting the best inventory policy for improving the performance of steel Industry.
In practical application, information about the inventory between probability distribution assumptions may be ambiguous to the experimenter and difficult to determine. To address this problem, our results are based on real data including sales and inventory quantity from an owner of the steel corporate. The whole data set covered the period from January, 2010 to February, 2014, a total of 216 data points. We also consider lead time, hold inventory and in-transit inventory constraints in ours system.
First, we use the information before 164 periods (1~164periods) to build the transition probability matrix of each action for all states in system. And through the conditions to calculus the optimal decision by MDP. Then, we use the optimal decision to test last 52 periods (165~216periods). The results show that, from the perspective of cost minimization, the best inventory policy would assign lower reorder quantity and the level of safety stock and also can meet demand for each period.
|
|---|
