Modified Two-Stage Degradation Model for Dynamic Maintenance Threshold Calculation Considering Uncertainty

Manufacturing systems are subject to unscheduled downtime due to machine malfunction. A threshold or an alarm is usually obtained based on equipment lifetime distribution to trigger maintenance work orders. The gap between conventional method and industrial practice is that the threshold is fixed and does not consider the status change due to machine degradation. In addition, the uncertainty involved in threshold calculation has not been carefully considered. An approach to extract dynamic maintenance threshold is presented for the continuously monitored degrading system considering system uncertainty. A modified two-stage degradation modeling approach is proposed to derive the degradation path for reliability prediction. A numerical case is exploited to demonstrate the benefits of the proposed approach. Results indicate that unnecessary or excessive maintenance is effectively avoided, and the machine availability will be consequently increased.