Session: Poster & Student Poster Session

Paper Number: 170352

170352 - Non-Destructive Rail Monitoring for Defect Identification 

Abstract: 

Non-destructive evaluation (NDE) of railway rails is essential to ensure safety in railway transportation. This study proposes a method for identifying noise sources and defects by using a reference database. The method involves constructing a reference dictionary containing noise correlation functions created over a localized area of the rail. The identification of the noise zone is based on calculating the mean square error (MSE) between the correlation for a defect and those in the dictionary.

Experimental tests were performed on a rail subjected to various forces applied at different positions. The reference dictionary was developed to include defect signatures for different force-position pairs. Advanced signal processing techniques, such as Coda Wave Interferometry (CWI), were used to mitigate temperature variations in the data, improving the accuracy of the results.

Preliminary results suggest that this method is promising for Structural Health Monitoring (SHM) of railway infrastructure, offering a reliable means of detecting defects. Additionally, deep neural networks are being tested to enhance the analysis and classification of defect signals. These networks are trained to increase fault detection accuracy and improve localization capabilities in rail systems. The integration of AI techniques and sparse methods further enhances the potential of this method for future rail monitoring applications.

Presenting Author: Lynda Chehami UPHF/LU

Presenting Author Biography: Associate Professor (HDR)
Member of CNU 63
President of the ULTRASOUNDS Technical Committee of the European Acoustics Association (EEA)
https://euracoustics.org/technical-committees/ultrasounds

Authors:

Elissa Akiki UPHF/LU
Lynda Chehami UPHF
Nikolay Smagin UPHF
Emmanuel Moulin UPHF
Jamal Assaad UPHF
Youssef Zaatar Lebanese University