Bayesian Approach for Fatigue Life Assessment of High-speed Trains Using In-service Monitoring Data

Abstract

This paper presents a method for fatigue life prediction of high-speed trains using in-service monitoring data, which establishes a Bayesian framework to take into account the uncertainties of the material parameters and in-service stresses imposed on the train. In this regard, a statistical treatment of fatigue test data and in-service stresses of the high-speed train components is required, which leads to probabilistic formulations of both the S-N relationship and stress spectrum. The formulated models allow the description of uncertainties in both the material parameters and in-service stress ranges from the posterior probability density functions (PDFs) updated by Bayesian inference. Subsequently, the information of these PDFs is utilized to develop a fully Bayesian procedure for predicting fatigue life of the high-speed train components. For verification, the proposed method is applied to predict the fatigue life of a train running on a China high-speed railway. Analytical solution for model derivations is provided. The probabilistic variance of the evaluated results of fatigue life in the Bayesian context demonstrates the influence of uncertainties inherent in the material properties and measurements.