Prognostics of polygonalization of high-speed railway train wheels using a generalized additive model smoothed by spline-backfitted kernel

Abstract

A method for the prognosis of polygonalization in high-speed railway train wheels is developed based on a generalized additive model. Unlike most previous studies, this study uses field data, so findings can help improve practical maintenance efficiency. A spline-backfitted kernel is used to improve computation efficiency when figuring out model parameters. This prognostics method can be applied to practical railway management decisions.