On-line possibilistic diagnosis based on expert knowledge for engine dyno test benches

Olivier de Mouzon (1) , Xavier Guérandel (2) , Didier Dubois (1) , Henri Prade (1)

1 IRIT-ADRIA - Argumentation, Décision, Raisonnement, Incertitude et Apprentissage
2 UT3 - Université Toulouse III - Paul Sabatier

Car engine electronic control unit tuning is a tedious and difficult task. Engineers can no more check on-line the measurements validity: Control strategies are becoming more complex and measurements more numerous. A fault diagnosis approach relying on experts’ qualitative and uncertain knowledge has been developed to identify whenever a malfunction is present. The theoretical frame-work - based on possibility theory and on consistency and abductive indices - is first explained. The latest results are then presented. Finally, the whole incremental development process of the prototypes is described, explaining the most difficult points and how they were handled.

Mots clés: diagnosis; engine dynamometer; uncertainty; fuzzy sets; possibility theory

Domaines: Intelligence artificielle [cs.AI]

Olivier de Mouzon, Xavier Guérandel, Didier Dubois, Henri Prade. On-line possibilistic diagnosis based on expert knowledge for engine dyno test benches. IFIP 18th World Computer Congress TC12 First International Conference on Artificial Intelligence Applications and Innovations (AIAI 2004), IFIP, Aug 2004, Toulouse, France. pp.435-448, ⟨10.1007/1-4020-8151-0_37⟩. ⟨hal-03370919⟩