This paper presents the results from a pilot study of a 'smart system' used for defect detection in railroad rails, particularly the critical transverse-type defects. The experimental data used to train the pattern recognition 'smart system' were extracted from experiments conducted during a previous long-range ultrasonic guided wave study conducted at the University of California, San Diego. Reflection coefficient plots corresponding to a variety of transverse and oblique defects were shown to provide features that were successfully used to train a 'smart system' to identify the defects automatically. This paper presents a brief introduction to support vector machines, followed by a description of the procedure used to determine the best data to be used to train the 'smart system', and concludes with lessons learned during this study.

Automatic defect classification in long-range ultrasonic rail inspection using a support vector machine-based ?smart system?

This report presents an ultrasonic method for defect detection in rails that substantially differs from conventional ultrasonic rail inspection. The main difference is in the use of ultrasonic guided waves traveling along the rail running direction. These waves have the potential for improved reliability in detecting transverse cracks and for improved inspection speed when compared to conventional rail inspections. Extensive experimental and numerical tests have been conducted to demonstrate the capabilities of guided waves for the detection of both small and large transverse cracks in the rail head. Advanced signal processing based on wavelet transforms and automatic pattern recognition has been implemented in the testing. Finally, impact hammers, pulsed lasers and air coupled sensors have been investigated as practical ways to perform the testing in the field.

On-Line High-Speed Rail Defect Detection

This paper analyzes support vector machines (SVMs) and several commonly used soft computing paradigms for pattern recognition including neural and wavelet networks, and fuzzy systems. Bayesian classifiers, fuzzy partitions, etc and tries to outline the similarities and differences among them. Support vector machines provide a new approach to the problem of pattern recognition with clear connections to the underlying statistical learning theory. We try to bring SVMs into the framework of the unification paradigm called the weighted radial basis function paradigm. Unifying different classes of methods has enormous advantages, such as the ability to merge all such techniques within the same system. It is hoped that this paper would provide theoretical guides for the study and applications of support vector machine and soft computing paradigms.© (2001) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.

Ying Li

Papers

Unification of support vector machines and soft computing paradigms for pattern recognition