Damage Detection of Train-Track System with Extreme Function Theory

Abstract

This paper focuses on detecting the damage of train–track system with the extreme function theory, which is a probabilistic method for damage detection. The train–track system adopted takes the coupling of vehicle and track into account. To detect the damage of the train–track system, the function of vehicle response with respect to the marching velocity of the train is used. The function can be represented by several discrete multi-dimensional data points. The extreme function theory deals with novelty as the distinction of the distribution tails of the conditional probability density. Normal model is constructed from the functions collected from vehicle in good service. A case study of train-track system with damage from the second suspension stiffness verifies the method well.