240 likes | 379 Views
Optimizing preventive maintenance for mechanical components using genetic algorithms. 指導教授:童超塵 老師 作者: Yuo-Tern Tsai, Kuo-Shong Wang, and Hwei-Yuan Teng 出處: Reliability Engineering and System Safety, Vol. 74, pp.89-97, 2001 報告人:陳建旻. Outline. Introduction
E N D
Optimizing preventive maintenance for mechanical components using genetic algorithms 指導教授:童超塵 老師 作者:Yuo-Tern Tsai, Kuo-Shong Wang, and Hwei-Yuan Teng 出處:Reliability Engineering and System Safety, Vol. 74, pp.89-97, 2001 報告人:陳建旻
Outline • Introduction • Effect of 1P-maintenance to reliability • Maintenance benefit analysis • Genetic algorithms • Case study • Conclusion
1. Introduction (1/3) • 維護分為:矯正維護(corrective maintenance)與預防維護(preventive maintenance)。 • 矯正維護可分為:小維修(minimal repair-1C)與矯正替換(corrective replacement-2C)階段。 • 預防維護可分為:預防維護(preventive maintenance-1P)與預防替換(preventive replacement-2P)階段。
1. Introduction (3/3) 一般而言,會將系統分為個別的零組件或單位進行維護,需考量其失效程度、成本、風險與使用壽命,排定預防維護階段,最後,利用基因演算法求得每單位成本下最大化的零組件壽命。
2. Effect of 1P-maintenance to reliability-2.1. Dynamic reliability (1/2) • Ao: initial reliability; A1: degraded factor • Ao與A1可經由模擬或實驗設計取得
2.1. Dynamic reliability (2/2) • mj-1: the improvement factor of 1P at the (j-1)-th PM stage • t: the effective age of component • tp: the maintenance interval
2.2. Improvement factor assessment • pij: the probability of 1P activities taken • dij: the improvement level
3. Maintenance benefit analysis (1/4) • Crj: the CM cost at the j-th stage • C0: the mean cost of per CM action of the system • Cp: the PM cost • Cpi: 1P and 2P cost on subscript p=1 and p=2
3. Maintenance benefit analysis (2/4) • Tj+1: the extended life of the system on the j-th stage • Bm: the unit-cost life of system • Cs0: the acquired cost of the system • Crn: the CM cost occurred in the last life stage (from tj to Tj+1)
3. Maintenance benefit analysis (3/4) • Bd: the discarded life of system
5. Case study-5.1. Problem formulation (1/3) • 電子式零組件失效多為隨機發生;機械式零組件失效則為多重損壞所致。
5.1. Problem formulation (2/3) • Rs(t): the system reliability • RC(t): the reliability of the surplus part of the system except the PM component
6. Conclusion • 經由量化的計算模式排定1P與2P預防維護階段,可適用於真實的系統。 • 利用動態可靠度(Dynamic reliability)的方程式結合基因演算法排定預防維護計畫,可有效減少計算時間。 • 當零組件的改善因子(improvement factor)愈高,則執行1P的機會將高於2P。