Analysis of Axial Acceleration for the Detection of Rail Squats in High-Speed Railways

Abstract

A squat is a type of fatigue defect caused by short-wavelength rotational contact; if squats are detected early, the maintenance cost of the track can be effectively reduced. In this paper, a method for the early detection of squats is presented based on ABA (axle box acceleration) and frequency signal processing techniques. To increase the measurement sensitivity for the squat, ABA was used to measure the longitudinal vibration. Compared to vertical ABA, longitudinal ABA does not include vibrations from rail fasteners and sleepers, so it is possible to effectively measure the vibration signal in relation to the impact of the rail. In this paper, vibration data were measured and analyzed by installing a 3-axis accelerometer on the wheel axle of the KTX; squat signals were more effectively extracted using the longitudinal vibration measurement presented above. The algorithm to detect the position of squats was developed based on wavelet spectrum analysis. This study was verified for the section of a domestic high-speed line, and as a result of conducting field verification for this section, squats were detected with a hit rate of about 88.2%. The main locations where the squats occurred were the rail welds and the joint section, and it was confirmed that unsupported sleepers occurred at locations where the squats occurred in some sections.