Title :
Considerations in implementing canary based prognostics
Author :
Mathew, Sony ; Osterman, Michael ; Pecht, Michael
Author_Institution :
Center for Advanced Life Cycle Engineering (CALCE) University of Maryland, College Park, MD 20742
Abstract :
The concept of using canary devices to predict the failure of a system is gaining momentum but there are practical issues that challenge the mainstream implementation of canary device based prognostics. This paper provides a general methodology for implementation of the canary based prognostics. This paper presents an approach to tackle practical issues including determining the number of canary devices required and the confidence in the prediction for a particular number of canaries. A failure prediction scheme to estimate system failure based on the failure of the canary device in the field, is also presented in this paper.
Keywords :
Loading; Mathematical model; Probability; Resistors; Shape; Temperature distribution; Canary design; failure prediction scheme; number of canaries; prediction confidence;
Conference_Titel :
Prognostics and Health Management (PHM), 2015 IEEE Conference on
Conference_Location :
Austin, TX, USA
DOI :
10.1109/ICPHM.2015.7245062