A New Fault Diagnosis Algorithm that Improves the Integration of Fault Detection and Isolation

Abstract

This work proposes a new model-based fault diagnosis method that improves the integration of the fault detection and isolation tasks. A new interface between fault detection and fault isolation is presented that contains information about the degree of fault signal activation and the occurrence time of fault signals. A combination of five fault signature matrices is used for the fault isolation process. The matrices store knowledge about faulty system behavior: boolean fault signal occurrence, signs of residual violation, sensitivities, time of fault signal activation and fault signal occurrence order. Finally, the new method is applied to the well-known two-tanks benchmark problem.