Track 2 Session 3
1:00 to 2:00 p.m. Tuesday June 9, 2009

A Maintenance Data-Driven Reliability Analysis and Knowledge Discovery Approach

Aircraft engines are required to maintain high levels of reliability, which are guaranteed by imposing stringent life limits on critical components. At the same time, large amounts of maintenance data are collected and could be used for accurate and actionable in-service reliability calculation for the different components of the system. However, this task is complicated by the large amount of data as well as competing risks and data censoring. "Competing risks" refers to the fact that each engine component may fail in a number of different ways while "data censoring" refers to the fact that critical components are often replaced or repaired before they fail as they reach their life limit. In this presentation, innovative reliability analysis techniques that are well suited to addressing the competing risk and data censoring problems are used to calculate component reliabilities. Knowledge discovery techniques are used to automate the data analysis in preparation for reliability calculation, to detect significant changes in reliability and to provide alerts to the user for further investigation and remediation.