A hybrid inspection-replacement policy for multi-stage degradation considering imperfect inspection with variable probabilities

Inspections are crucial to reveal asset health and support preventive maintenance, which, however, are susceptible to multiple uncertainties due to human errors and limitations of diagnosing technologies. This often leads to inspection errors with variable probabilities affected by operational time/state, particularly for multistage degradation. This paper investigates a hybrid inspection-replacement policy for two-stage continuously degrading assets subject to time-state-variant inspection errors, either false positive or false negative. The error probabilities upon each inspection are related to (a) the accumulated operational age, and (b) whether encounters defects. To mitigate loss due to errors, a hybrid replacement planning integrating three types of replacements is scheduled. In particular, age-centered replacement is implemented to relieve uncertainty interference of inspection errors. Additionally, threshold-based replacement (when degradation attains a threshold) and defect-induced replacement (when the asset is reported as defective) are executed to ensure timely response to state variations. The long run cost rate is minimized through the joint optimization of the inspection interval, degradation threshold and age limit. The applicability of the proposed model is demonstrated through numerical experiments conducted on operations & maintenance management of high-speed train bearing, which is proved to be cost-effective than several comparative models.

Automated summary

This paper investigates a hybrid inspection-replacement policy for two-stage continuously degrading assets subject to time-state-variant inspection errors. To mitigate loss due to errors, a hybrid replacement planning integrating three types of replacements is scheduled. The long run cost rate is minimized through the joint optimization of the inspection interval, degradation threshold and age limit.