Study of Signal Processing Techniques for Detecting Rail Corrugation Using Axle-Box Accelerometers in a Scaled Experimental Track

Abstract

Using a scaled experimental track, this paper demonstrates that the detection of corrugation is possible via the axle-box acceleration (ABA) measurements and adequate signal processing technique. To conduct the quantitative analysis of the effect of the wavelength and amplitude of the corrugation on the vehicle vibration, a corrugated profile is designed as a combination of four harmonic waves and machined on the railhead by using a CNC machine. It is proved that vertical ABA signals can produce more reliable results for the identification of corrugation wavelength and the exact corrugation location, than the longitudinal ABA signals and the vertical bogie accelerations. In addition, a normalized vertical ABA is defined and found to be a good choice for the corrugation estimation, due to the velocity-independent nature of the amplitudes. Consequently, a methodology is developed to determine the wavelength and the amplitude of the corrugation based on the ABA spectrum and the transfer function of the wheel-rail system. The experimental results with different vehicle speeds show that the newly developed approach leads to the accurate results for the longer wavelengths of the corrugation.

Keywords: Vehicle vibration, Normalized acceleration, Transfer function, Correlation, time-frequency analysis