Condition-based Maintenance for Performance Degradation under Non-periodic Unreliable Inspections

With the development of monitoring technology, increasing attention has been paid to condition-based maintenance (CBM). Also, if condition information for maintenance can be obtained with fewer inspections, the cost of the entire maintenance process could be reduced. However, in application, the monitoring equipment cannot maintain always a reliable operating state, for example, due to various uncertain factors, such as sensor errors, component tolerances, and environment disturbances, the inspections from the sensors are often unreliable. Motivated by these simple observations, we propose a nonperiodic CBM policy under unreliable inspections. The time interval of the nonperiodic inspections is obtained via an inspection scheduling function. The unknown parameters of the component degradation process are updated by GD, while simultaneously the maintenance decision variables are adjusted. A catastrophe strategy-based particle swarm optimization is used to set the optimal decision variables by minimizing the long-run cost rate. Application to laser degradation data illustrates the effectiveness of the proposed method.