Maintenance decision-making model based on POMDP for traction power supply equipment and its application

Abstract

Traditional annual maintenance planning of traction power supply equipment relies mostly on old convention and subjective experience, having the problem of excessive maintenance and insufficient maintenance. Now, the application of Prognostics and Health Management (PHM) theory is gradually implemented, condition-based maintenance (CBM) has become the first choice for the traction power supply equipment maintenance mode. In view of the decision-making problems we encountered when the maintenance mode of traction power supply equipment transforms to the condition-based maintenance, an annual state inspection and maintenance plan decision-making model of the traction power supply equipment was established based on Partially Observable Markov Decision Process (POMDP). The model considers the uncertainty in the health state assessment, and equipment health state transferring regularity is obtained based on historical data. The equipment synthetic cost per unit time throughout the year is set as the optimization goal, and One-Pass algorithm was adopted to achieve the optimization of the traction power supply equipment's annual maintenance plan. The results show that the model is feasible and effective, providing basis to implement scientific and reasonable condition-based maintenance schemes for the traction power supply equipment.