Numerous maintenance policies have been proposed in the reliability mathematics and engineering literature. Nevertheless, little has been reported on their practical applications in industries. This gap is largely due to restrictive assumptions of the maintenance policies. Two of the main assumptions are that maintenance is conducted on typical components and that the reliability of an item under maintenance is known (where the item can be a component or a system composed of multiple components). These assumptions do not often hold in the real world: maintenance is often performed on a collection of components such as an integrated circuit plate and the reliability of each individual component may not be known. To reduce these gaps, this paper develops a new maintenance policy for a collection of components and an approximate method to estimate the reliability of this collection based on the failure data collected from the field. The maintenance policy considers that a system is composed of three subsystems with different levels of maintenance effectiveness (i.e, minimal, imperfect, and perfect). The approximate estimate of the reliability of each subsystem is derived based on the failure data that are time between failures of the system but not those of the components that cause the system to fail. An algorithm for simulating the superposition of generalised renewal processes is then proposed. Numerical examples are used to illustrate the proposed approximation method.

中文翻译：

---

预防性维护策略和近似多组件系统故障过程的方法


可靠性数学和工程文献中已经提出了许多维护策略。然而，关于它们在工业中的实际应用的报道却很少。这种差距主要是由于维护政策的限制性假设造成的。两个主要假设是对典型组件进行维护，并且维护中的项目的可靠性是已知的（其中该项目可以是一个组件或由多个组件组成的系统）。这些假设在现实世界中通常并不成立：维护通常是对一组组件（例如集成电路板）进行的，并且每个单独组件的可靠性可能未知。为了缩小这些差距，本文为一组组件开发了一种新的维护策略，并根据从现场收集的故障数据来估计该组组件的可靠性的近似方法。维护策略认为系统由具有不同维护有效性级别（即最小、不完美和完美）的三个子系统组成。每个子系统可靠性的近似估计是根据故障数据得出的，这些故障数据是系统故障之间的时间，而不是导致系统故障的组件的故障数据。然后提出了一种模拟广义更新过程叠加的算法。数值例子用于说明所提出的近似方法。