Toward Information System Architecture to Support Predictive Maintenance Approach
Résumé
The prognostic and health management (PHM) approach aims at supporting maintenance operations in order to ensure the functionality of a system. In order to achieve this objective, a PHM approach is composed of a prognostic component, able to send a prognostic of failure, and a component able to give the health status of the system. Nowadays, this approach suffers from a lack of exploitation of the emerging technologies. This article presents a novel architecture for PHM approach able to extract added value from data. This lambda architecture embeds two layers: a speed layer and a storage layer. Thanks to the storage layer, maintenance rules can be applied as well as the result of machine learning algorithms to the speed layer in order to realize the prognostic aspect of the PHM. In addition, the system has to deal with heterogeneous data, which comes with the necessity to handle the big data issues as well as making it interoperable. This is achieved thanks to a service-oriented architecture approach and the use of complex event processing.