Pattern Recognition Method for Off-Board Automotive Vehicle Failure Isolation
| Authors: | Charkaoui Nasser, Université de Technologie de Compiègne/PSA Peugeot Citroën, France
Dubuisson Bernard, Université de Technologie de Compiègne, France
Ambroise Christophe, Université de Technologie de Compiègne, France
Boatas Armand, PSA Peugeot Citroën, France |
|---|
| Topic: | 6.4 Safeprocess |
|---|
| Session: | Applications of Fault Diagnosis and Fault Tolerant Control |
|---|
| Keywords: | Diagnosis, pattern recognition, classification, binary elements, decision theory, automobile industry, failure detection and isolation. |
|---|
Abstract
This paper introduces a pattern recognition method specially dedicatedto the detection and isolation of after-sale failures on an automotive vehicle. Thismethod is well suited to deal with binary data. It is based on the similaritybetween an input pattern and some reference patterns characteristic of each failureclass. The initial space representation is divided into several subspaces in orderto allow classification of patterns which belong simultaneously to several classes.Experiments with real data demonstrate the good performance and the adaptedstructure of this classification system.
