IMPROVED FLAGGING FOR PATTERN CLASSIFYING DIAGNOSTIC SYSTEMS

Fault detection and isolation (diagnosis) is based on residual generat ion and residual analysis. The model-based approach flags the residual s through thresholding, to isolate the effect of faults from noise, an d performs diagnosis by mapping the residuals to a residual space with prespecified fault signatures. The main problem with this approach is that thresholds are not always able to differentiate between the effe ct of faults and noise, so this approach suffers from false alarms, un detected faults, and misdiagnosis. As an alternative to prespecified f ault signatures and to cope with their variability, the use of pattern classification techniques has been proposed. However, since the fault signatures established by these classifiers are formed irrespective o f diagnosability, this approach is also prone to misdiagnosis. In this paper we demonstrate the application of a flagging unit that enhances the quality of fault signatures. This Unit, which relies on a trainin g set to tune its parameters, is shown to improve detection, reduce th e number of false alarms and enhance diagnostics.