Fault estimation using bicausal bond graph and linear fractional transformation

Abstract

The problem of safety and reliability of complex systems motivates the development of new fault detection, isolation and estimation approaches. Indeed, this paper propose a new methodology of fault estimation based on the bond graph tool and on its causal and structural proprieties. Using the linear fractional transformation, a new procedure of fault modeling is proposed on the graphical model. The bi-causality notion is used on the graphical model in order to deduce systematically the mathematical fault estimation equations. The developed methodology is applied and validated by experimental data of an electromechanical system.