skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Fault detection and isolation for linear time-invariant systems

Abstract

This paper is concerned with the problem of detecting and isolating faults by an observer. If the initial error of state estimation is zero, we show that faults can be isolated if and only if the system has a left-invertible detectability matrix which is defined in the present paper. For the more realistic case of nonzero initial error of state estimation, we develop fault isolation filters such that failures to be detected can be asymptotically isolated. We give necessary and sufficient conditions for the existence of such filters. We provide a method to design a fault isolation filter. This design procedure guarantees to isolate up to n faults, where n is the dimension of the system.

Authors:
;  [1]
  1. Arizona State Univ., Tempe, AZ (United States)
Publication Date:
OSTI Identifier:
70853
Report Number(s):
CONF-941216-
CNN: Contract RP8015-03; TRN: 95:004350-0040
Resource Type:
Conference
Resource Relation:
Conference: 33. Institute of Electrical and Electronic Engineers (IEEE) conference on decision and control, Orlando, FL (United States), 14-16 Dec 1994; Other Information: PBD: 1994; Related Information: Is Part Of Proceedings of the 33rd IEEE conference on decision and control; PB: 4347 p.
Country of Publication:
United States
Language:
English
Subject:
22 NUCLEAR REACTOR TECHNOLOGY; 42 ENGINEERING NOT INCLUDED IN OTHER CATEGORIES; FAILURES; DETECTION; CONTROL SYSTEMS; DESIGN; SYSTEM FAILURE ANALYSIS; AVAILABILITY; RELIABILITY

Citation Formats

Liu, B., and Si, J.. Fault detection and isolation for linear time-invariant systems. United States: N. p., 1994. Web.
Liu, B., & Si, J.. Fault detection and isolation for linear time-invariant systems. United States.
Liu, B., and Si, J.. Sat . "Fault detection and isolation for linear time-invariant systems". United States. doi:.
@article{osti_70853,
title = {Fault detection and isolation for linear time-invariant systems},
author = {Liu, B. and Si, J.},
abstractNote = {This paper is concerned with the problem of detecting and isolating faults by an observer. If the initial error of state estimation is zero, we show that faults can be isolated if and only if the system has a left-invertible detectability matrix which is defined in the present paper. For the more realistic case of nonzero initial error of state estimation, we develop fault isolation filters such that failures to be detected can be asymptotically isolated. We give necessary and sufficient conditions for the existence of such filters. We provide a method to design a fault isolation filter. This design procedure guarantees to isolate up to n faults, where n is the dimension of the system.},
doi = {},
journal = {},
number = ,
volume = ,
place = {United States},
year = {Sat Dec 31 00:00:00 EST 1994},
month = {Sat Dec 31 00:00:00 EST 1994}
}

Conference:
Other availability
Please see Document Availability for additional information on obtaining the full-text document. Library patrons may search WorldCat to identify libraries that hold this conference proceeding.

Save / Share:
  • Consider the time system Ex(k + 1) = Ax(k) + Bu(k) where E is a singular square matrix. It is assumed that the system is either a priori regular i is regularizable by a feedback law of the form u(k) = Ky(k) + V(k). or it The problem is this: Find an input sequence which will drive x(k) from a given x(0) to a desired {open_quotes}final vector{close_quotes} x(N) in a given number of steps N while minimizing the cost. The novelty of this paper`s approach is the use of singular-value decomposition and of weighted generalized inverses.
  • The similarities and differences of a universal network to normal neural networks are outlined. The description and application of a universal network is discussed by showing how a simple linear system is modeled by normal techniques and by universal network techniques. A full implementation of the universal network as universal process modeling software on a dedicated computer system at EBR-II is described and example results are presented. It is concluded that the universal network provides different feature recognition capabilities than a neural network and that the universal network can provide extremely fast, accurate, and fault-tolerant estimation, validation, and replacement ofmore » signals in a real system.« less
  • The similarities and differences of a universal network to normal neural networks are outlined. The description and application of a universal network is discussed by showing how a simple linear system is modeled by normal techniques and by universal network techniques. A full implementation of the universal network as universal process modeling software on a dedicated computer system at EBR-II is described and example results are presented. It is concluded that the universal network provides different feature recognition capabilities than a neural network and that the universal network can provide extremely fast, accurate, and fault-tolerant estimation, validation, and replacement ofmore » signals in a real system.« less
  • In this paper a methodology for instrument data validation as well as fault detection and isolation, based on analytic redundancy, is presented. This work differs from previously reported work on analytic redundancy in that validation of all the main parameters of the plant heat transport loops is sought by using plant-wide instrument information. An LMFBR plant is used as a reference, and validation of the following plant parameters is considered: reactor power (Q), reactor inlet (T/sub IC/) and reactor outlet (T/sub OC/) coolant temperatures, intermediate heat exchanger (IHX) inlet (T/sub IS/) and outlet (T/sub OS/) secondary coolant temperatures, steam generatormore » feedwater temperature (T/sub w/), steam temperature (T/sub s/) and pressure (P/sub s/), as well as primary (G/sub p/), intermediate (G/sub I/), and feedwater (G/sub w/) flow. In this paper, only validation at steady state conditions will be discussed.« less
  • In this paper we discuss two advanced techniques, process fault detection and nonlinear time series analysis, and apply them to the analysis of vector-valued and single-valued time-series data. We investigate model-based process fault detection methods for analyzing simulated, multivariate, time-series data from a three-tank system. The model-predictions are compared with simulated measurements of the same variables to form residual vectors that are tested for the presence of faults (possible diversions in safeguards terminology). We evaluate two methods, testing all individual residuals with a univariate z-score and testing all variables simultaneously with the Mahalanobis distance, for their ability to detect lossmore » of material from two different leak scenarios from the three-tank system: a leak without and with replacement of the lost volume. Nonlinear time-series analysis tools were compared with the linear methods popularized by Box and Jenkins. We compare prediction results using three nonlinear and two linear modeling methods on each of six simulated time series: two nonlinear and four linear. The nonlinear methods performed better at predicting the nonlinear time series and did as well as the linear methods at predicting the linear values.« less