多種需求分配下備份件補給系統的存貨配置

Abstract

隨著國防安全議題的日趨受到重視，軍方武器系統若於演練操作或實戰時期能夠有效運作將是非常重要，而負責維持系統設備正常運作的備份件則是其關鍵所在，若備份件存量過少，便無法適時對失效的系統進行維修、更換，因而耽誤生產作業；若備份件存量過多時，則會導致庫存成本過高，故如何提供一有效的備份件補給體係實是一重要的研究課題。

然而，過去備份件的相關文獻並無在多階補給架構與多階系統結構，以及有限維修能力之下，針對不符從卜瓦松失效分配行為的零件進行最佳化存貨配置的研究。故本研究提出一個較佳化求解模型，首先乃藉商用軟體OPUS10取得一起始解；再者利用eM-Plant模擬平台建構一具考量多種需求分配之能力的評估程式；最後結合邊際配置法進行較佳化作業，並發展4個方法針對邊際演算法進行修正，以強化求解效率。而在X系統的個案中，不僅說明零件失效行為並不完全符從卜瓦松分配模式，更在備份件存貨配置上證明確實存在高達10%的成本節省空間。

The spare-part stocking problem is very important for weapon system. Keeping a higher inventory level would increase weapon’s availability, yet at the price of incurring higher costs. Most related literature assumed that the failure of a spare part follows a Poisson distribution. This would reduce the problem complexity; yet may not well model a particular real world problem. This research aims to solve the spare-part stocking problem in a scenario where the failures of spare parts are not limited to Poisson distribution. The proposed method involves three major steps. First, we use OPUS10, proprietary software, to yield an initial solution. Second, we use simulation technique to evaluate the performance of a stocking solution. Third, we develop some heuristics to efficiently and effectively improve the solution until a satisfactory one is obtained. Experiment results show that the proposed method as compared to OPUS10 may save about 10% in inventory cost.