Authors:
Y. Kourd
1
;
N. Guersi
2
and
D. Lefebvre
3
Affiliations:
1
Faculty of Science and Engineer, Mohamed Khider Biskra University, Algeria
;
2
Université Badji Mokhtar Annaba, Algeria
;
3
Greah, Université Le Havre, France
Keyword(s):
Fault Diagnosis, Modelling, Residual Generation, Residual Evaluation, Neural Classifier, Neurofuzzy Classifiers.
Related
Ontology
Subjects/Areas/Topics:
Control and Supervision Systems
;
Informatics in Control, Automation and Robotics
;
Robotics and Automation
;
Surveillance
Abstract:
Fault diagnosis is a major challenge for complex systems as long as it increases the safety and productivity. This work concerns faults diagnosis, based on artificial intelligence, neural networks, and fuzzy logic. Thanks to an associative memory, neural networks have good capacities of organization, approximation and classification. Combined with fuzzy logic, neural networks are an effective tool for system modelling, fault detection and fault diagnosis. This paper illustrates the potential of these tools for the modelling and the diagnosis of an industrial actuator (DAMADICS benchmark).