An inspection-based replacement planning in consideration of state-driven imperfect inspections

Inspection, as a fundamental component of asset management, is crucial to reveal underlying defects and report health evolution. Due to technical restrictions or human errors, inspections are often imperfect to miss the actual system state, posing challenges to the optimality of decision-makings. This study investigates an inspection-based replacement management strategy for two-phase continuous deteriorating systems subject to inspection errors. The non-steady evolution trajectory is captured by piecewise stochastic process with random state transition time. Inspections are equally spaced to reveal the underlying system state (both discrete and continuous), with a certain probability to miss the hidden defective state. Upon inspection, two levels of control limits are jointly scheduled to adjust execution frequencies of preventive replacement. The cost model is formulated and optimized, and the impact of imperfect inspection on system performance is assessed. The applicability of the proposed optimization model is validated through a case study from high-speed train bearings, which scales well in operational cost control, reducing about 39 and 4% cost compared with two conventional policies.