Optimal perfect corrective maintenance policy for a system with multiple components using data‐driven decision‐making methods

Abstract

AbstractThis paper considers a multicomponent series system to be inspected periodically. It is assumed that the exact cause of system failure might be unknown, and we only know that it belongs to a subset of all components, called masked set. In this case, it is said that the exact cause of failure is masked. A perfect corrective maintenance policy is applied to repair the system, such that a failed component is replaced by a new one at the first inspection time after failure. The distance between each two inspections is considered a decision parameter that should be optimized. Three levels of information availability are considered, including full, partial, and noninformation availability, in which the maintenance policy should be optimized. The former is considered a benchmark, and the statistical and machine learning data‐driven algorithms are used to estimate the cost for the other two information levels. Eventually, using Monte Carlo simulation studies, the proposed method application is analyzed.