Adaptive diagnosis observer for dynamical systems with dual actuator faults: A sign-corrected regressor design

Abstract

Safe critical systems often require punctual and accurate faults alarming and diagnosis to facilitate subsequent coping and handling. In this paper, an adaptive observers-based detection, isolation and diagnosis algorithms are designed for dynamical systems with dual frequently encountered types of actuator faults, i.e., additive and multiplicative faults. Fault detector, capable of circumventing false fault alarming in the presence of external disturbance and false fault relieving signals, is firstly designed, generating punctual faults detection signals. Followed by an adaptive diagnosis observer using novel sign-corrected regressor, rapid isolation and accurate diagnosis are thus guaranteed. The highlight of proposed sign-corrected regressor based diagnosis algorithms lies in that unbiased faults estimations are ensured even with systematic inputs without persistency of excitation. The convergence of adaptive diagnosis observer is guaranteed in the sense of Kalman–Yakubovich–Popov lemma, and state estimation and diagnosis error dynamics is ensured to be SPR. Robustness limit against external disturbances in the form of analytical expression is given. Simulation examples are presented to verify the effectiveness of theoretical findings.