A new robust fault diagnosis approach based on bond graph method

Abstract

In this paper, a new robust fault diagnosis approach based on the combination of Bond Graph (BG) method and a robust observer is proposed. Different criteria including disturbance attenuation level, enhanced fault sensitivity and robustness against parametric uncertainties are considered in the proposed fault diagnosis system. A new form of analytical redundancy relations (ARRs), which is called error-based analytical redundancy relations (EARRs), is introduced and is used as the residuals for fault diagnosis purposes in considered uncertain linear systems. This form of the residuals is obtained using the output estimation error of the observer and satisfies considered criteria of the fault diagnosis system. The ARRs and state space representation of the system can be derived based on the BG model. The robust observer design is defined as an LMI problem in which the region of robustness against parametric uncertainty is achieved by the feasibility constraint of the obtained LMI. Finally, in order to show the performance of the proposed method, a quarter-car model is considered and different criteria are compared with the conventional ARRs method.