Air-coupled ultrasonic assessment of concrete rail ties

• Published in: NDT & E international : independent nondestructive testing and evaluation 2021-10, Vol.123, p.102511, Article 102511
• Main Authors: Evani, Sai Kalyan, Spalvier, Agustin, Popovics, John S.
• Format: Article
• Language: English
• Subjects: Contactless; Crosstie; Evaluation; f-k domain; Machine learning; NDT; Nondestructive; Support vector machine
• ISSN: 0963-8695; 1879-1174
• DOI: 10.1016/j.ndteint.2021.102511

This paper investigates a contactless non-destructive evaluation technology for fast and efficient condition assessment of concrete rail ties. Ultrasonic surface waves are generated in rail ties using an air-coupled ultrasonic transmitter and are sensed at another location along the tie using an array of micro-electromechanical system (MEMS) sensors. Concrete rail ties with different types of damage are evaluated. The measured signal data are visualized in time-space (t-x) and frequency-wavenumber (f-k) domains, and four signal parameters are derived. Signal parameters measured from the damaged ties are compared with those from undamaged/healthy ties to generate two- and four-dimensional decision spaces. Threshold decision boundaries separating healthy and damaged rail ties in the decision space are computed using the support vector machine (SVM) algorithm, and the ability of these decision spaces to distinguish between healthy and damaged rail ties is investigated using statistical data analysis. The results demonstrate that high quality signal data are obtained from concrete rail ties using the proposed air-coupled configuration. The four-dimensional decision space, which uses all four signal parameters together, can distinguish between healthy and damaged rail ties with accuracies of around 80 %.