Title :
False Alarm Mitigation of Vibration Diagnostic Systems
Author :
Byington, Carl S. ; Watson, Matthew J. ; Amin, Sanket ; Begin, Michael
Author_Institution :
Impact Technol. LLC, Rochester, NY
Abstract :
False alarms in legacy aircraft diagnostic systems have negatively impacted fleet maintenance costs and mission readiness. As the industry moves towards more advanced prognostic and health management (PHM) solutions, a reduction in false alarms is needed to reduce the cost and readiness burdens that have plagued legacy systems. It is therefore important to understand why these false alarms occur and how they are generated so appropriate mitigation solutions can be included in next-generation diagnostic systems. This paper examines four major sources of false alarms in the development of vibration diagnostics (faulty sensor performance, transient system operating conditions, improper health indicator selection, and inadequate fault detection logic) and details a solution designed to mitigate their impact. An overview of the developed false alarm statistics toolbox for PHM (FAST PHMtrade) software is also provided to illustrate how the software guides design engineers through the processes of verifying data, processing for diagnostic features, analyzing feature performances, developing "virtual" features through fusion, and deriving statistically optimized feature thresholds. The developed approach will improve the overall performance, robustness, and reliability of vibration diagnostic and prognostics systems.
Keywords :
aerospace computing; aircraft maintenance; condition monitoring; fault diagnosis; reliability; PHM software; false alarm mitigation; false alarm statistics toolbox; fault detection logic; fleet maintenance costs; health indicator selection; health management; legacy aircraft diagnostic systems; mission readiness; next-generation diagnostic systems; transient system operating conditions; vibration diagnostic reliability; vibration diagnostic systems; Aircraft; Costs; Fault detection; Logic design; Prognostics and health management; Sensor systems; Software design; Software performance; Software tools; Statistical analysis;
Conference_Titel :
Aerospace Conference, 2008 IEEE
Conference_Location :
Big Sky, MT
Print_ISBN :
978-1-4244-1487-1
Electronic_ISBN :
1095-323X
DOI :
10.1109/AERO.2008.4526620