Optimization of Condition-based Maintenance for Traction Power Supply Equipment based on Partially Observable Semi-Markov Decision Process

Due to uncertainty of state evaluation, the result of state evaluation may not be consistent with the actual state of the equipment. A maintenance scheme ignoring this uncertainty can cause additional downtime and economic costs. Aiming at this problem, a condition-based maintenance (CBM) model is established based on the partially observable Markov decision process (POMDP) which is continuous in time. The continuous-time Markov chain is used to describe the degradation process of traction power supply equipment (TPSE). The equipment state is classified into four levels and the transition probabilities between different states are solved. In order to quantify the uncertainty of state evaluation, the state-observation probability is introduced. Considering the maintenance cost and the failure risk, the optimal maintenance method and inspection period are determined based on this model. Finally, the recorded degradation data of 27.5-kV vacuum circuit breakers for a traction power supply system (TPSS) are used in the numerical example to illustrate the effectiveness of this model.