Assessment of a Robust Algorithm for the Detection and Diagnosis of Additive Faults

This paper addresses the problem of additive faults (such as input/output sensor and actuator) in a dynamic system, from the view of system identification techniques. The relation between the residuals of the model–based fault diagnosis and the innovations of the system identification procedure is implemented and corresponding algorithms are extracted for the tracking of additive faults, while robustness to noise and disturbances is issued. The study is initiated using single input-single output models and extended to multiple inputs-multiple outputs structures. Furthermore, the detection problem of additive faults for systems with unobservable excitation is examined.