Probabilistic model forecasting for rail wear in seoul metro based on bayesian theory

Abstract

A safe and reliable railway operation requires an organic and systematic approach to railway maintenance. Despite the importance of timely and valid track maintenance and applicability of inspected data to the optimum track management process, inspected wear data inspected by a railway inspection system in Korea have not been utilized for decision making of maintenance scenario, but just accumulated. Moreover, the process of inspecting wear data includes some uncertainties, probabilistic-based models have more reasonable application in field. This can be accomplished by developing probabilistic-based stochastic model considering uncertainties for the prediction of rail wear using inspected data. This paper reports on the development and verification of a probabilistic forecasting model for rail wear progress. This developed forecasting model utilizes the particle filter method concept based on Bayesian theory and real inspected wear data of Seoul Metro are applied to verify the model.