Contribution à la gestion de la maintenance dans un environnement distribué

Abstract

Nowadays technological developments and the current economic context have brought about important changes in the maintenance management systems, the choice of its structural elements is not independent of its environment. In parallel with technological developments, the spatial and distributed organization of maintenance has also been developed. In this context, we have proposed a maintenance management system applied to a production unit consisting of several geographically distributed sites. To have well-maintained and reliable facilities capable of making estimates faster and more reliable, a well-designed maintenance program is required. To do this, we applied a predictive maintenance action implemented in CBM. A fault detection and isolation method in the time-frequency domain on an industrial bearing is applied. The experimental results obtained are validated with respect to the results calculated theoretically. The generated predictive maintenance tasks are required by random events, assigning these tasks to resources with different skill levels. This problem is reduced to a dynamic scheduling problem with independent parallel resources this problem is solved by applying neighborhood search heuristics based on an improved descent algorithm. In order to show the efficiency of our management system and as our system meets the standards of object-oriented design (COO), we have developed an object-oriented model with the UML language. Simulation and validation of the maintenance management system is performed.