Plausible reasoning theory in reliability-centered maintenance analysis

Abstract

Reliability-centered maintenance (RCM) is a powerful method for choosing optimal maintenance tactics. By itself, though, the method misses opportunities to deploy expert-level knowledge to the maintenance people performing the analysis. Existing software approaches that attempt to fill this gap, such as expert systems, have numerous drawbacks and in some cases can be detrimental, causing the user to develop a false sense of security. Plausible reasoning theory suggests ways that evidence can be gathered and used efficiently to draw conclusions about maintenance tactics. In conjunction, Bayesian belief networks can be used to capture in software the reasoning strategy suggested by plausible reasoning theory. The author created prototype software using these theories and approaches. The resulting software represents a model of RCM logic and allows the user to work with the model to produce effective results. During the entire analysis process, the user can see instantly the effect that various inputs have on the resulting recommendations. Users are able to pursue the most promising paths of reasoning, arriving at the best RCM solution quickly.