Design of an innovative and self-adaptive-smart algorithm to investigate the structural integrity of a rail track using Rayleigh waves emitted and sensed by a fully non-contact laser transduction system

Abstract

The focus of this study is on locating surface and sub-surface defects that occur in rail tracks using Rayleigh waves that were emitted and sensed by a fully non-contact laser transduction system. As the quality of received signals varies with respect to the rail surface characteristics, spotting the reflection from a defect can be extremely challenging. These signals are in general contaminated with noise and have low repeatability that could hinder the proper identification of the defect signal. In view of this, an innovative signal processing technique called a self-adaptive-smart algorithm (SASA) was designed and developed. In SASA, the incident wave that is the first coming wave-packet is taken as a mother wavelet. A library of possible mother wavelets was designed based on the experimental data. As the incident wave for each sensing point varies because of the physical condition of the rail surface and the laser excitation, the algorithm automatically picks the mother wavelet from the library that generates the largest absolute cross-correlation with the incident wave. SASA is found to be able to suppress the unwanted wave packets from the overall signal leaving behind only the incident wave for a healthy specimen, and the incident wave and its reflection from the defect for a damaged specimen. The functioning of the algorithm was successfully tested by carrying out extensive experiments on a real rail track in the presence of different types of surface and sub-surface defects on its head and web. The obtained results prove the effectiveness of using SASA in localizing defects in rails with a marginal error. Notably, the proposed method has benefits such as being self-adaptive, can help suppress high levels of noise, bring the peak of defect reflected wave in the center, and distinguish between a healthy and damaged sample.