Real-time flaw characterization through learning-by-examples techniques: a comparative study applied to ECT

Shamim Ahmed 1 Roberto Miorelli 1 Marco Salucci 2, 3 Andrea Massa 2, 3
1 LSME - Laboratoire de Simulation et de Modélisation Électromagnetique
DISC - Département Imagerie et Simulation pour le Contrôle : DRT/LIST/DISC
2 L2S - Laboratoire des signaux et systèmes
3 ELEDIA - ELEDIA Research Center@DISI, University of Trento, via Sommarive 5, 38123 Trento, Italy
Abstract : In this paper, real time flaw characterization inside the structure under test (SUT) exploiting Learning by Examples (LBE) strategies is illustrated in the framework of Electromagnetic NDT (E-NDT). In particular, several LBE methodologies are investigated when dealing with the inversion of Eddy Current Testing (ECT) data. The performance of LBE algorithms is evaluated in terms of CPU time efficiency, and prediction accuracy on noisy data. Towards this end, a comparison between different regression techniques is shown by considering both synthetic and real experimental ECT data.
Type de document :
Communication dans un congrès
21th International Workshop on Electromagnetic Nondestructive Evaluation (ENDE’16) , Sep 2016, Lisbon, Portugal. 2016
Liste complète des métadonnées
https://hal.archives-ouvertes.fr/hal-01774236
Contributeur : Andrea Massa <>
Soumis le : lundi 23 avril 2018 - 14:56:45
Dernière modification le : mercredi 23 janvier 2019 - 14:39:31

Identifiants

  • HAL Id : hal-01774236, version 1

Collections

CEA | CEA-UPSAY | DRT | CENTRALESUPELEC | LIST | SUP_LSS | L2S | UNIV-PARIS-SACLAY | SUP_SIGNAUX

Citation

Shamim Ahmed, Roberto Miorelli, Marco Salucci, Andrea Massa. Real-time flaw characterization through learning-by-examples techniques: a comparative study applied to ECT. 21th International Workshop on Electromagnetic Nondestructive Evaluation (ENDE’16) , Sep 2016, Lisbon, Portugal. 2016. 〈hal-01774236〉

Exporter

BibTeX TEI DC DCterms EndNote

Partager

Métriques

Consultations de la notice

128

Contact

support.ccsd.cnrs.fr
hal.support@ccsd.cnrs.fr
Logo CCSD