| Summary: | 博士 === 國立成功大學 === 工業與資訊管理學系碩博士班 === 99 === Since a complex system usually deteriorates with age, and such deterioration may cause malfunctions and result in severe damage and losses, preventive maintenance (PM) is often carried out to keep the system functioning in a good state. The decision of when and how to perform a PM activity is commonly based on system age. However, when the deterioration is highly correlated with some physical (e.g., material wear or lubricant degradation) or system performance variables, such as quality of produced items or number of system breakdowns, a condition-based PM policy seems more appropriate than an age-dependent one. In this study, the deterioration of the system is modeled by a non-homogeneous Poisson process with a power law failure intensity, and a widely used age reduction model is applied to describe the restoration degree of imperfect PM activities. We derive the optimal non-periodic PM schedules which minimize the expected total cost per unit time for two popular and highly attractive PM policies in the literature: the age-dependent and the condition-based PM policies. However, since the determination of such the optimal PM schedules may involve numerous uncertainties which typically make the analyses difficult to perform because of the scarcity of data, a Bayesian decision model, which utilizes all available information effectively, is also proposed in the age-dependent PM context to deal with such difficulties. Numerical examples are given to illustrate the importance and the effectiveness of the proposed models. Sensitivity analyses and discussion on the optimal PM strategies for the proposed models are also presented.
|
|---|
