An algorithmic approach to fault diagnosis in linear systems

An algorithmic approach for multiple fault diagnosis of linear discrete-time systems is proposed. Based on notions of an expected deviation vector and variation factors, it is shown that t faults in functional units of a dynamic system can be diagnosed with t+1 sample times. The method is considered efficient when the number of faults is unknown but small and when the sampling period is lengthy, as in chemical process with large time constants. Its effectiveness is illustrated by simulated examples