Rail Wear and Remaining Life Prediction Using Meta-Models

The study presented in this paper proposes a method to estimate the Remaining Useful Life (RUL) of railway tracks determined by wear and taking into account various track geometry and usage profile parameters. The relation between these parameters and rail wear is established by means of meta-models derived from physical models. These models are obtained with regression analysis where the best fit is found from a relatively large set of numerical experiments for various scenarios. The specific parameter settings for these scenarios are obtained by using the Latin Hypercube Sampling (LHS) method. Furthermore, for the rail profile, which is one of the input parameters for the meta-model, it is shown that the evolution due to wear in moderate curves can be characterized by only one parameter. The findings in this work including are valuable for Infrastructure Managers (IMs) and can easily be implemented in maintenance decision support tools.