Predictive maintenance optimization for manufacturing systems considering perfect and imperfect inspections: application to injection molding machine

Predictive Maintenance is a promising strategy for anticipating failures of manufacturing systems through degradation control. While inspections assess health and guide maintenance decisions, their costs pose challenges. To minimize inspection costs, this paper proposes an adaptive non-periodic inspection model combining cost-effective (but less reliable) imperfect inspections with accurate (yet costly) perfect ones. The originality of this approach lies in its ability to dynamically optimize inspection types and intervals based on system lifetime prognosis. The model is validated through a case study of an injection molding machine, demonstrating superior performances compared to conventional inspection models.