Improving Real Time Rail Monitoring with EMAT Sensors and Superlets Analysis

Abstract The development of online and in filed ultrasonic testing tools for rail defects detection requires advanced signal processing analysis methodologies. The objective is to properly identify the Time of Flight (ToF) of the wave in high noise measurements. We propose an estimator of this physical parameter with the use of a novel time-frequency analysis technique, adapted from the Continuous Wavelet Transform (CWT), called the Superlet Transform (SLT). It provides higher accuracy even at low Signal to Noise Ratio (SNR). We apply this estimator on a simulated example to quantify the reached performances, in terms of errors and variances.