Skip to Main content Skip to Navigation
Theses

Observation et diagnostic de processus industriels à modèle non linéaire : application aux machines électriques

Abstract : This thesis focuses on the definition of a robust strategy for the diagnosis of industrial processes with nonlinear model. The defined strategy is based on the use of nonlinear observers not only for diagnosis but also for control of these systems. The aim is threefold. The synthesized observer will reconstruct the state variables, will be sensitive to faults for diagnosis purpose while being robust to disturbances and parametric uncertainties for control purpose. Two observers were studied for this matter. The first observer is a Kalman-like observer. This observer has been applied to detect multiplicative faults for a DC motor series. The stability of the observer for the control and the diagnosis has been proven for two cases of parameters faults. The second observer is a High Gain observer. It has been applied to stator short-circuits fault diagnosis for induction machines. The High Gain observer is used for the diagnosis of induction machine, with and without mechanical sensor. The performance of fault detection algorithms for induction motor has been evaluated on a specific benchmark "Observer for the Diagnosis" defined by the working group Inter GDR CE2. This benchmark is located at IRCCyN.
Document type :
Theses
Complete list of metadatas

Cited literature [101 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-00676588
Contributor : Alain Glumineau <>
Submitted on : Monday, March 5, 2012 - 4:26:43 PM
Last modification on : Wednesday, December 19, 2018 - 3:02:04 PM
Long-term archiving on: : Friday, November 23, 2012 - 3:55:50 PM

Identifiers

  • HAL Id : tel-00676588, version 1

Collections

Citation

Ayan Mahamoud. Observation et diagnostic de processus industriels à modèle non linéaire : application aux machines électriques. Automatique / Robotique. Ecole Centrale de Nantes (ECN), 2010. Français. ⟨tel-00676588⟩

Share

Metrics

Record views

1054

Files downloads

14149