Longitudinal Degradation Data Simulation Method Based on Sequential Statistics

In order to trace the degradation process of individual railroad wagon wheels for effective preventive differential maintenance, a longitudinal data generation method is proposed applying to wheel tread wear data that are highly aggregated and ungrouped. A generalized linear model with a Gamma linkage function is used to model the overall degradation of the wheel tread wear process, and the variation space of individual wheel tread wear increments at adjacent inspection mileage is constructed by sequential statistics to trace the wheel wear trajectory. Comparison with experimental wheel tread wear data shows that the proposed method can effectively generate longitudinal data, which could provide ideas for individual wheel degradation modeling and differential maintenance.