DocumentCode
183921
Title
Component failure estimation with Unknown Input Observer: A physical approach
Author
Gahlouz, I. ; Tarasov, E. ; Sueur, C. ; Bouamama, B. Ould
Author_Institution
LAGIS, Polytech Lille, Villeneuve-d´Ascq, France
fYear
2014
fDate
8-10 Oct. 2014
Firstpage
390
Lastpage
396
Abstract
In the literature Unknown Input Observers for sensor and actuator failure estimation are widely developed. None of them concerns component fault estimation. These analytical approaches are based on state equation description and complex numerical calculation. The present paper deals with physical component failure estimation based on bond graph approach as power full methodology for dynamic modeling of multi-physical systems. The additional energy due to the component failure (clearly displayed in the model) is considered as unknown input. Furthermore, properties of this observer are based on the finite and infinite structures and are synthesized directly from the graphical model, which can be pointed out from physical, structural and causal properties of used graphical representation. The developed methodology is validated in real experimental hydraulic process allowing valve and tank leakage failure estimation.
Keywords
actuators; bond graphs; failure analysis; numerical analysis; observers; sensors; actuator failure estimation; bond graph approach; complex numerical calculation; component failure estimation; component fault estimation; experimental hydraulic process; graphical model; graphical representation; infinite structures; multiphysical systems; physical approach; sensor failure estimation; state equation description; tank leakage failure estimation; unknown input observers; valve leakage failure estimation; Detectors; Equations; Mathematical model; Observers; Valves; Vectors; Bond graph approach; Component fault modeling and estimation; Unknown input observer;
fLanguage
English
Publisher
ieee
Conference_Titel
Control Applications (CCA), 2014 IEEE Conference on
Conference_Location
Juan Les Antibes
Type
conf
DOI
10.1109/CCA.2014.6981377
Filename
6981377
Link To Document