Optimized database for training neural networks used in non-destructive testing

Issue title: Proceedings of the twelfth International Symposium on Interdisciplinary Electromagnetic, Mechanic and Biomedical Problems, ISEM Bad Gastein

Article type: Research Article

Authors: Gyimóthy, Szabolcs^{a; *} | Bihan, Yann Le^b | Pávó, József^a

Affiliations: [a] Department of Broadband Infocommunications and Electromagnetic Theory, Budapest University of Technology and Economics, Egry J. u. 18, H-1521 Budapest, Hungary | [b] Laboratoire de Génie Electrique de Paris, CNRS – Supélec – UPS – UPMC, 11 rue Joliot-Curie, 91192 Gif-sur-Yvette Cedex, France

Correspondence: [*] Corresponding author. Tel.: +36 1 463 2918; Fax: +36 1 463 3189; E-mail: [email protected]

Abstract: Neural network is applied for the non-destructive evaluation of in-material defects based on the measured or simulated data of an eddy-current testing experiment. The network is trained with the data measured for suitably selected defect prototypes, so that this selection constitute a consistent representation of the forward problem. This network performs in defect reconstruction better, than those networks trained with randomly or regularly selected defect prototypes of about the same number.

Keywords: Inverse problems, neural networks

DOI: 10.3233/JAE-2007-766

Journal: International Journal of Applied Electromagnetics and Mechanics, vol. 25, no. 1-4, pp. 717-721, 2007