Constraint suspension is a physical model based approach to fault detection and diagnosis. Its capabilities are demonstrated by describing how it might be applied to data recorded from an air conditioning unit installed in a factory. One possible implementation is outlined and results obtained from applying this are given. Issues relating to suitability, accuracy and sensitivity are discussed. Constraint suspension is unlikely to be readily applicable to the detection of faults embedded in controller implementations and it is therefore proposed that a simple rule-based system be appended to accommodate this limitation.
Get full access to this article
View all access options for this article.
References
1.
Davis R.Diagnostic reasoning based on structure and behaviourArtificial Intelligence24347-410 (1984)
2.
Holman J.P.Heat Transfer5th edn ( New York : McGraw Hill) 1985.
3.
deKleer J. and Williams B.C.Diagnosing multiple faultsArtificial Intelligence3297-130 (1987)
4.
Reiter R.A theory of diagnosis from first principlesArtificial Intelligence3257-95 (1987)
5.
Howell J.Model-based fault detection in information poor plantsAutomatica30(6) 929-943 (June 1994 )
6.
Leary J.J. and P.J. GawthropProcess fault detection using constraint suspensionIEE 1'roc. Part D134(4) 264-271 (1987)
7.
Gawthrop P.J. and N.A. MarrisonFault detection, location and identification in dynamic systemsProc. European Control Conf.21911-1917 (1991)
8.
Maddison E.J.An application for constraint suspension: fauk diagnosis in an air conditioning unit MSc thesis, University of Glasgow ( 1994)
9.
Jones P.Air conditioning engineering3rd edn ( London: Arnold) ( 1985)
