Application des Réseaux de Neurones pour la Caractérisation Géométrique d’un Défaut par Courants de Foucault

Abstract

In this work, we have presented different nondestructive testing methods. Than, the steps of finite element modeling of eddy current nondestructive testing are explained. In order to achieve the reconstruction of defect profile while knowing the sensor impedance and position, we have used neural networks method. After the training step, the obtained profile is found to be largely similar to that of the actual one.