An approach for joint scheduling of production and predictive maintenance activities

Abstract

The Industry 4.0 paradigm, thanks to the deployment of cutting-edge technologies enabling the deployment of new services, contributes to improve the agility of productive organizations. Among these services, the Prognostic and Health Management (PHM) contributes to the health assessment of the manufacturing resources and to prognose their future conditions by providing decision supports for production and predictive maintenance management. However, the future conditions of technical production resources depend on the productive tasks they will have to carry out. If their future conditions will not satisfy production criteria, maintenance tasks will have to be planned and productive tasks will be delayed or assigned to other resources for which their future conditions considering these new tasks must be assessed. In this context, a multi-agent system SCEMP (Supervisor, Customers, Environment, Maintainers and Producers) is here proposed in which production scheduling and predictive maintenance planning collaborate and exploit decision supports provided by PHM modules. The proposed multi-agent system provides a framework in which production and the predictive maintenance activities can be scheduled simultaneously by compromising on their objectives. During the scheduling process, SCEMP enables to identify the needed predictive maintenance from the assignments of production tasks to machines, the machine component prognoses and machine models. It schedules production tasks and predictive maintenance activities according to the number, competencies and availabilities of production and maintenance resources. The SCEMP framework is described and presented in the tough job shop context. For this context, case studies have been generated and scheduled within acceptable computation times. To illustrate the SCEMP functioning, some case studies are detailed with the obtained performances. It is flexible and can be adapted to various manufacturing situations. It can also be used to assess the interest of implementing prognostic functions for machine components.