根據維護相關文獻,針對有限時間域的局部週期預防維護模型,當給定維護次數,較短的維護間隔時間能得到較佳的期望總成本。然而,當給定維護次數時,相關文獻均假設維護間隔時間被限制在某一範圍內,而無法小於該範圍。因此,本研究為獲得更好的預防維護政策,乃放鬆維護間隔時間的限制條件,在給定維護次數時,期能找到更小的維護間隔時間,並期望獲得比文獻更小的期望總維護成本。此外,新設備在使用初期不易發生失效。依逐次型預防維護政策特性,設備自全新至第一次的預防維護間隔時間較其他預防維護間隔時間長,且逐次型預防維護政策的期望總維護成本會優於週期型。然而,週期型預防維護比逐次型預防維護在管理上或實務上易於執行。因此,本研究乃結合逐次型與週期型預防維護模型的優點,假設展延週期型預防維護政策之起始時間能獲得比一般週期型預防維護政策較低的期望總維護成本。綜合以上的分析,以期望總維護成本最小化為目標函數,發展兩種週期預防維護修正模型以及最佳解的演算法,並找出最佳的週期型預防維護政策以及優於文獻中相對應的原始模型最佳預防維護政策的條件。研究結果顯示,修正模型在特定條件下能優於文獻中的原始模型,且所推導的最佳解相關定理亦與範例驗證的結果相符。
Due to rapid development of science and technology, it makes production equipment more accurate and complex, setup cost and maintenance cost more expensive. Besides, in order to reduce unpredictable the possibility of failure occurrence and to maintain systems in a steady state and a high level of reliability, industries have focus on equipments or systems maintenance issue. According to the literature, it is found that a shorter interval between each PM can result in a better expected total maintenance cost (TC) in the partially periodical preventive maintenance models over a finite time span. The optimal policy of the preventive maintenance (PM) models from the literature (called the original PM model) is obtained by searching PM interval over the specified range [Tmin,Tmax) for a given number of PM. In order to obtain smaller TC than the Orignal PM model, we release the constraint of the searching range of the PM interval. From another perspective, due to new equipments usually will not fail easily in initial stage of the useful life, we consider the idea of postponing the starting time of the maintenance policy. In this research two assumptions are provided, (1) shorten the PM interval and (2) postpone the starting time of the maintenance policy. Based on the assumptions, two modified periodical preventive maintenance models are developed and the algorithms of finding the optimal policies by minimizing the expected total maintenance cost are proposed. Finally, we verify the optimal policies of two types of the modified PM models with the examples and show the conditions that the proposed optimal PM policies are better than their corresponding similar PM models.