|
Activity Number:
|
334
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 8, 2006 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Physical and Engineering Sciences
|
| Abstract - #306938 |
|
Title:
|
Modeling Unit Degradation from Exceedance Time Data
|
|
Author(s):
|
Hui Fan*+ and Brock Osborn and Thomas R. Willemain and Pasquale Sullo
|
|
Companies:
|
Rensselaer Polytechnic Institute and GE Global Research and Rensselaer Polytechnic Institute and Rensselaer Polytechnic Institute
|
|
Address:
|
CII 5015, DSES Department, troy, NY, 12180,
|
|
Keywords:
|
degradation ; exceedance ; jet engine ; reliability
|
|
Abstract:
|
Monitoring and predicting equipment degradation is an important task within reliability analysis. We consider the problem of exploiting degradation data censored to reveal only the times at which a performance variable exceeds a threshold. Our motivating example is the study of exhaust gas temperatures in jet engines, where safety regulations define exceedance thresholds and require expensive off-wing maintenance after a specified number of exceedances. Using both artificial data from a random walk model and real data, we develop methods for using data on exceedance times to estimate unit-specific degradation parameters and develop predictive distributions for the time until a specified number of exceedances will occur. We also assess the use of nonparametric methods to test for the presence of performance degradation.
|
