Fault Estimation and Isolation Using Bond Graph Approach

This paper deals with fault detection and isolation using bond graph approach. The latter consists of using the analytical redundancy relations to detect the fault, and the fault signature matrix to isolate it. The generation of analytical redundancy relation is performed using the causality inversion method to automate the procedure. The contribution of this paper is the development of an algorithm of fault isolation for the faults which have the same signature. The developed procedure is based on the residuals sensitivity and in the generation of the fault estimation equations. The fault is estimated from each residual sensible to it and compared with the others to isolate it. An application on a mobile robot is done to validate this procedure.