2013
IMAACA 2013 Proceeding
Bond graph proportional-integral observer-based robust fault detection
Authors: Ghada Saoudi, Rafika EL Harabi, Geneviève Dauphin-Tanguy, Belkacem Ould Bouamama, Mohamed Naceur Abdelkrim
Abstract
The present paper investigates a bond graph tool to design full-order proportional-integral (PI) observers for a robust fault detection purpose. The proposed method allows the calculation of the gain matrix graphically through covering causal paths and loops based on the pole placement techniques for linear systems. The robust residuals are further generated from an uncertain bond graph model in linear fractional transformation (LFT) form so as to detect actuator faults in presence of parameter uncertainties. Simulation tests on a hydraulic system show the dynamic behavior of system variables and the robustness of the PI observers in the presence of modeling errors. The effectiveness of the proposed robust fault detection estimator is later illustrated via a DC motor.
I3M Scientific Sponsors
I3M Industrial Sponsors
I3M Media Sponsors
