Optimizing a maintenance strategy by combining age-based maintenance and an imperfect fault detection model

Abstract

In predictive maintenance, fault detection models are often imperfect, with both false positives and false negatives. By solely planning maintenance based on an imperfect fault detection model, the number of failures, maintenance tasks and maintenance costs might increase compared to age-based maintenance. In this paper, we therefore combine age-based maintenance with an imperfect fault detection model. First, we derive the probability of a false positive given the age of the component, using Bayes theorem. We subsequently use the renewal reward theory to optimize the age at which to preventively replace components, if we also replace components based on the alarms of the fault detection model. Here, we may also ignore alarms when the component is just installed. The optimal parameters of this approach indicate whether it is optimal to perform preventive maintenance based on i) only the age (age-based maintenance), ii) only the alarms of the fault detection model, iii) based on the alarms and the age or iv) to have no preventive replacements. We illustrate this approach in a small case study, where we show that even fault detection models with many false positives and false negatives can lower the maintenance costs, if combined with age-based replacements.