Abstract

Fault diagnosis aims to diagnose system failures and to enable timely maintenance that, in turn, can minimize system maintenance costs. In order to evaluate and maximize the benefits from fault diagnosis, the life‐cycle maintenance cost should be analyzed. This paper presents a framework for life‐cycle maintenance cost analysis that considers time‐dependent false and missed alarms in fault diagnosis. First, time‐dependent false and missed alarms are proposed. The false and missed alarm rates are not static but vary depending on the health state of the engineered system, which changes over time. Second, a life‐cycle maintenance cost analysis framework is proposed. This is based upon a stochastic simulation method that can incorporate time‐dependent false and missed alarm rates and various uncertainties, such as health degradation and health restoration through maintenance. Third, a fault diagnosis model design method is proposed, based upon the proposed life‐cycle maintenance cost analysis framework. The proposed method incorporates optimal false and missed alarm weights into the life‐cycle maintenance cost. As a result, the proposed ideas enable accurate estimation and minimization of the overall life‐cycle maintenance cost. The effectiveness of the proposed methods is demonstrated via a numerical example and an electro‐hydrostatic actuator case study.