Fault Diagnosis System Based on Ontology for Fleet Case Reused
Résumé
Maintenance plays a key role by improving system availability, performance efficiency, and product quality. Condition based maintenance plus (CBM+) and Prognostics and Health Management (PHM) maintenance strategies propose new maintenance approaches in a “predict and prevent” view. In these anticipative approaches, early diagnosis plays a key role. Such a diagnosis is hard since only partial information is available. We propose to use the benefit of past event occurred on similar systems to help diagnosis. The originality of the approach lies in the consideration of systems that are not identical to the one under study. Indeed, similar systems are considered in order to gather more relevant and wider information to handle current diagnosis case. The level of similarity is controlled using an ontology in order to broaden or narrow the search. The chapter proposes first a state of the art of the uses of a fleet for PHM strategies and more specifically for diagnosis. Secondly, the fleet-case-reused is presented within a feedback cycle in order to control the resulting information. Thirdly the ontology engineering and the embedded knowledge are described. Finally, using an application in the naval domain, the resulting software is presented and three scenarios show the advantages of the proposed approach.