Using Bayesian belief networks to predict the reliability of military vehicles

Abstract

Predicting the reliability of military vehicles has traditionally concentrated on estimation using failure data gathered during trials or use. However, it is increasingly recognised that predicting reliability earlier in the life cycle, using design and process capability evidence, is one way of improving predictions and positively influencing reliability. This article presents the use of Bayesian belief networks (BBNs) as a decision support tool to achieve these twin goals. The BBN models presented are built into the TRACS software tool, which is in daily use within DERA Land Systems.