1) Research center: Centre de Recherche en Automatique de Nancy
The «Centre de Recherche en Automatique de Nancy» (Automatic Control) is a Research Centre funded by the " Centre National de la Recherche Scientifique (CNRS) " and two universities in Nancy: UHP (Université Henri Poincaré) and INPL (Institut National Polytechnique de Lorraine). (http://www.cran.uhp-nancy.fr) has well established activity covering 30 years of research in the field of Fault Detection and Isolation (FDI) in uncertain and complex control systems for industrial control problems (jet engines, satellite attitude control, steel mills, environmental process, etc).
2) Title: Fault diagnosis of nonlinear system by kernel principal component analysis.
3) Description: In the general sense, a fault is something that changes the behavior of a system such that it does no longer satisfy its purpose. Therefore fault diagnosis is of great practical significance whatever the system being considered. It involves to detect faults, to isolate the faulty components and to characterize faults.
Principal component analysis (PCA) has been applied successfully in the monitoring of complex systems. PCA is a widely used method for dimensionality reduction when there exists some data redundancy that stems from linear relation between process variables. Unfortunately for nonlinear system, linear PCA is unable to disclose nonlinear relationships between variables. This has naturally motivated various developments of nonlinear PCA in an effort to model non trivial data structures more faithfully. Compared to other nonlinear methods based on neural networks,
the kernel-PCA method (k-PCA) involved an interesting formalism for learning k-PCA model.
Therefore the goal of this project is to explore k-PCA method for fault detection and isolation.
Firstly future work consists in generalizing the methods developed for fault diagnosis by linear PCA to k-PCA. The project is expected to make contributions to all aspects of fault diagnostic, from k-PCA model learning theory to fault isolation.
4) Requirement: The candidate will have a PhD in applied mathematics or engineering (process control and monitoring, ...). Experience in machine learning or related field (with publications) is required. Applications should comprise a letter of motivation, a CV plus the names and contact information of two individuals who have agreed to send letters of recommendation.
5) Duration: 12 months, starting in January 2010.
6) Salary: Equivalent to a starting French assistant professor (around 28 k\euro/year) (including coverage of health insurance).
7) Advisor: José Ragot
Adresse : CRAN, 2 avenue de la Forêt de Haye, F-54506 Vandoeuvre.
E-mail : jose.ragot@ensem.inpl-nancy.fr
Téléphone : 03 83 59 56 82 - Fax: 03 83 59 56 44
web : http://www.ensem.inpl-nancy.fr/Jose.Ragot
Please quote 10 Academic Resources Daily in your application to this opportunity!
