Train Wheel Rim Degradation Modeling based on Generalized Linear Mixed Effect Model

The wear of the train wheel rim is one of the main failure mode of railway train wheel, which could be view as a slow degradation process. Degradation modeling of rim is the foundation of remaining useful life prediction and maintenance decision optimization. In this paper, a generalized linear mixed effect model is proposed to model a non-linear degradation shape with over-dispersion and individual random effects. The results show that the proposed model is effective to deal with non-normal error, non-homogenous variance, and non-linear path degradation data.

I. Introduction

The government of China is vigorously deducing the high-quality development of the equipment manufacturing industry. Rail transit manufacturing is an important industry of "Made in China 2025" and an important support for the "Belt and Road". For the consideration of system reliability, economy, and safety, Prognostics and Health Management (PHM) has become a research hotspot in the field of reliability. Modeling the degradation process under intermittent monitoring is one of the cores and basic contents of PHM [1].