Bayesian-driven predictive replacement planning in consideration of spare parts ordering

This paper proposes a Bayesian-driven joint optimization policy of predictive replacement and spare parts ordering, where crucial lifetime parameters are updated at inspections to support joint planning. The health evolution trajectory is modeled by a generalized-form Wiener process, whose crucial pattern parameters are estimated combining Bayesian inference and maximum likelihood estimation (MLE) algorithm. Furthermore, the real-time remaining lifetime is calculated upon each inspection to jointly update the optimal ordering time, predictive replacement time, through the setting of a dynamic reliability threshold to trigger subsequent replacement/ordering decisions. The feasibility and superiority of the proposed planning approach are verified through a practical case study on health management of train bearings.