Process Diagnosis in Transient Operating Regions Fault detection and isolation in a liquid sodium cooling system

Diagnosis of dynamic physical systems is com plex and requires close interaction of monitor ing fault generation and re nement and pre diction We have developed a methodology for model based diagnosis of abrupt faults in con tinuous systems that combines dynamic sys tem modeling with a qualitative reasoning ap proach A temporal causal system model is used to identify faults from deviant measure ments and predict future behavior expressed as signatures i e qualitative magnitude changes and higher order time derivative e ects A comparison of the transient characteristics of the observed variables with the predicted ef fects helps re ne initial fault hypotheses This allows for quick fault isolation and circum vents di culties associated with interactions caused by feedback and dependent faults This methodology has been successfully applied to the secondary cooling loop of fast breeder re actors Results on a representative set of sim ulated faults are discussed in this paper