Rapid detection of cracks in the rail foot by ultrasonic B-scan imaging using a shear horizontal guided wave electromagnetic acoustic transducer

Abstract Transverse cracks in the rail foot are responsible for the breaking of the rail. The existing non-destructive testing technology for high-speed railway track is difficult to perform in-situ and rapid detection for the rail foot, which poses a hidden danger to driving safety. In this study, a new method is proposed for the in-situ and rapid detection of cracks in the rail foot by ultrasonic B-scan imaging. A finite element method (FEM) based on the Bloch–Floquet boundary and domain constraint were employed to calculate the dispersion curves of shear horizontal (SH)-like guided waves propagating in the rail foot. SH guided waves mainly vibrate with in-plane displacement, and they are less affected by the rail components such as elastic clips and pads. Based on the calculated dispersion curves, an SH-guided wave electromagnetic acoustic transducer (EMAT) with a center frequency of 0.154 MHz was developed, and ultrasonic B-scan imaging was performed to detect straight cracks in the rail foot. Besides, the synchrosqueezed wavelet transform (SWT) method was proposed to remove noises and higher-order guided wave modes from the original ultrasonically detected signals. Results show that FEM with the Bloch–Floquet boundary and domain constraint can be used to solve the dispersion curves of the SH-like guided waves of the rail foot accurately and easily. The application of SWT can increase the lift-off of the EMAT, eliminate the noise, remove the higher-order SH-like guided wave modes, improve the detection efficiency, and enhance the quality of ultrasonic B-scan images. After the SWT processing, the signal-to-noise ratio (SNR) of the ultrasonic signal is increased by at least 5.98 dB. When the lift-off of the EMAT is 4 mm and without synchronous averaging, the on-line and rapid ultrasonic B-scan imaging of the rail foot using guided wave EMAT can still achieve a reasonable SNR.