Set‐membership based distributed fault detection and isolation for networked linear systems with disturbances

Abstract This paper investigates principles and algorithms for detection and isolation of different types of faults in networked discrete‐time linear systems with disturbances bounded by polytopes. Based on set‐membership methods to estimate system states, a novel distributed detection method using zonotopes and over‐approximation of reachable sets is proposed to enhance the sensitivity of detecting faults. For fault isolation, the distributed local measurements are utilized to quickly identify the location of the fault source. The effectiveness of the detection and isolation procedures are shown through theoretical analysis and numerical examples.