Time–frequency characterization of rail corrugation under a combined auto-regressive and matched filter scheme

Rail corrugation is an oscillatory mechanical wear of rail surface raising from the long-term interaction between rail and wheel. Signal processing approaches to corrugation monitoring, as recommended by the European standards for instance, are designed either in the mileage domain or in the wavelength domain. However a joint mileage and wavelength domain analysis of the monitoring data can provide crucial information about the simultaneous amplitude and wavelength modulations of the corrugation modes. It is proposed in this paper to perform such a mileage–wavelength domain analysis of rail corrugation using the class of Auto-Regressive-MAtched Filterbank (AR-MAFI) methods. We show that these methods assume a statistical model that fits the corrugation data. We discuss also the optimal parameter settings for the analysis of corrugation data. Experimental studies performed on data collected from the French RATP metro network show that the AR-MAFI methods outperform (in terms of readability and accuracy) the standard distance domain or wavelength domain methods in localizing and characterizing corrugation.

► Non-stationarity in rail corrugation patterns. ► Time–frequency analysis for diagnosis of rail corrugation. ► Rail corrugation mode tracking. ► Amplitude and wavelength estimation of rail corrugation modes.