An investigation of periodic degradation of axle box vibration spectrum for a high-speed rail vehicle based on Bayesian method

Axle box vibration serves as the main source of excitations for rail vehicles. Due to the wear of wheel/rail contact and the re-profiling procedure, the axle box vibration usually degrades periodically with the increased mileage in the service. This could significantly impact the estimation of vibration fatigue when the component is subjected to the axle box vibration. This paper develop a method to describe the periodic evolution of axle box vibration spectrum to better characterise the vibration spectrum of axle box. In this study, a Wiener process incorporating with four random parameters was employed to model the non-linearity of the degradation process. The maximum likelihood estimation (MLE) algorithm is used to estimate the initial values of the random parameters, and a Bayesian approach is employed to update the parameters based on newly obtained data. Finally, the proposed methodology is tested using long-term field test data from a high-speed train, and the results demonstrate that it accurately estimates the evolution of the axle box acceleration spectral density (ASD) spectrum. This could aid in predicting the residual service life of structures subjected to axle box vibration and further contribute to the development of maintenance strategies and top-down design of the structure.