Fuzzy modeling of analytical redundancy for sensor failure detection

Conference · Tue Jan 01 00:00:00 EST 1991 · Transactions of the American Nuclear Society; (United States)

OSTI ID:5765199

Tsai, T M; Chou, H P ^[1]

---

1. National Tsing Hua Univ., Hsinchu (Taiwan)

Failure detection and isolation (FDI) in dynamic systems may be accomplished by testing the consistency of the system via analytically redundant relations. The redundant relation is basically a mathematical model relating system inputs and dissimilar sensor outputs from which information is extracted and subsequently examined for the presence of failure signatures. Performance of the approach is often jeopardized by inherent modeling error and noise interference. To mitigate such effects, techniques such as Kalman filtering, auto-regression-moving-average (ARMA) modeling in conjunction with probability tests are often employed. These conventional techniques treat the stochastic nature of uncertainties in a deterministic manner to generate best-estimated model and sensor outputs by minimizing uncertainties. In this paper, the authors present a different approach by treating the effect of uncertainties with fuzzy numbers. Coefficients in redundant relations derived from first-principle physical models are considered as fuzzy parameters and on-line updated according to system behaviors. Failure detection is accomplished by examining the possibility that a sensor signal occurred in an estimated fuzzy domain. To facilitate failure isolation, individual FDI monitors are designed for each interested sensor.

• OSTI ID: 5765199
• Report Number(s): CONF-910603-; CODEN: TANSA
• Journal Information: Transactions of the American Nuclear Society; (United States), Vol. 63; Conference: Annual meeting of the American Nuclear Society (ANS), Orlando, FL (United States), 2-6 Jun 1991; ISSN 0003-018X
• Country of Publication: United States
• Language: English

Related Subjects

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS
99 GENERAL AND MISCELLANEOUS//MATHEMATICS, COMPUTING, AND INFORMATION SCIENCE
PWR TYPE REACTORS
REACTOR MONITORING SYSTEMS
DIAGNOSTIC TECHNIQUES
MATHEMATICAL MODELS
DETECTION
DYNAMICS
ERRORS
FAILURES
LEVEL INDICATORS
MITIGATION
NOISE
PERFORMANCE
PRESSURE MEASUREMENT
PRESSURIZERS
REACTOR COMPONENTS
SAMPLING
SIGNAL-TO-NOISE RATIO
TEMPERATURE MONITORING
ENRICHED URANIUM REACTORS
MEASURING INSTRUMENTS
MECHANICS
MONITORING
POWER REACTORS
REACTORS
THERMAL REACTORS
WATER COOLED REACTORS
WATER MODERATED REACTORS
210200* - Power Reactors
Nonbreeding
Light-Water Moderated
Nonboiling Water Cooled
990200 - Mathematics & Computers