Robust Diagnosis of Industrial Systems by Bond graph Model

Abstract

This paper aims to solve a research problem by a robust diagnosis of a hydraulic system with sand filter by approach of graph of connection using fractional linear transformations (BG-LFT). The method we develop is based on the use of analytic redundant relations by a bond graph graph model (BG-ARRs). These relationships not only allow the detection and isolation of defects on the various elements of the system, but also the location by structural and causal analysis. The results suggest that the use of the link diagram model for a valve fault (eg an R1 valve is blocked), figures 9.a), 9.b) and 9.c) show that the residual models r1 and r2 become non-zero, so the flow and pressure levels are zero at reservoir C2. These values mean that these residues are sensitive to the variation of the flow rate at the level of the valve R1, which is confirmed by the theoretical results presented in table 2. The simulation of the system is carried out by the software dedicated to the bond graph approach