Fast Fault Estimation and Fault Tolerant Control using Adaptive Observer

This paper presents, an Active Fault Tolerant Control (AFTC) based on combination of adaptive observer and Bond Graph (BG) method for linear systems. In the first stage, an adaptive observer is designed to detect and estimate failures. The observer gains are determined by solving a Linear Matrix Inequality (LMI). Next, once the fault has been isolated and its size estimated, the fault can be accommodated by creating a new control into the faulty system. In this context, fault tolerant control systems are involved in automation engineering issues as a remedy to system reliability. Thereby, we interest in this work to the Pseudo Inverse Method (PIM) law to compensate actuator fault effects in order to guarantee desirable closed-loop performances. The system modeling and fault diagnosis are designed by a graphical concept, the bond graph tool as a useful methodology for multidisciplinary systems and which is based on causal and behavioral properties. Finally, an hydraulic system with two-tank is considered and controlled to illustrate the efficiency of this approach.