NASA Logo

NTRS

NTRS - NASA Technical Reports Server

Back to Results
Detecting Faults By Use Of Hidden Markov ModelsFrequency of false alarms reduced. Faults in complicated dynamic system (e.g., antenna-aiming system, telecommunication network, or human heart) detected automatically by method of automated, continuous monitoring. Obtains time-series data by sampling multiple sensor outputs at discrete intervals of t and processes data via algorithm determining whether system in normal or faulty state. Algorithm implements, among other things, hidden first-order temporal Markov model of states of system. Mathematical model of dynamics of system not needed. Present method is "prior" method mentioned in "Improved Hidden-Markov-Model Method of Detecting Faults" (NPO-18982).
Document ID
19950065310
Document Type
Other - NASA Tech Brief
Authors
Smyth, Padhraic J. (Caltech)
Date Acquired
August 17, 2013
Publication Date
March 1, 1995
Publication Information
Publication: NASA Tech Briefs
Volume: 19
Issue: 3
ISSN: 0145-319X
Subject Category
MATHEMATICS AND INFORMATION SCIENCES
Report/Patent Number
NPO-18946
Distribution Limits
Public
Copyright
Work of the US Gov. Public Use Permitted.