Fault detection and diagnosis from a hybrid system point of view

Guoxi Feng , Mohamed Djemai (1) , Thierry-Marie Guerra (1) , Krishna Busawon

1 LAMIH - Laboratoire d'Automatique, de Mécanique et d'Informatique industrielles et Humaines - UMR 8201

This paper gives an analysis of fault detection and diagnosis (FDD) from a hybrid system point of view. A simple RL circuit is used to illustrate detectability and modelling issues in traditional observer-based fault detection approaches. We show that traditional model-based FDD whereby a faulty model of the system is sought is not valid especially in the case of abrupt faults. We show that a more appropriate way to tackle the FDD problem is to derive a hybrid model of the system based on the fact that the occurrence of an abrupt fault will send the system from one mode (not necessarily safe) to another faulty mode. We also discuss the issue of fault sensitivity with respect to an output and show sits relation to fault detectability. Simulation results are presented to support all the arguments put forward.

Mots clés: Fault detection; Observers; Electrical fault detection; Integrated circuit modeling; Sensitivity; Resistance

Domaines: Automatique / Robotique

Guoxi Feng, Mohamed Djemai, Thierry-Marie Guerra, Krishna Busawon. Fault detection and diagnosis from a hybrid system point of view. 2016 4th International Symposium on Environmental Friendly Energies and Applications (EFEA), Sep 2016, Belgrade, Serbia. pp.1-6, ⟨10.1109/EFEA.2016.7748778⟩. ⟨hal-03416898⟩