%0 Conference Proceedings %T On a Bayesian inversion approach in eddy-current testing %+ Laboratoire des signaux et systèmes (L2S) %+ Laboratoire Génie électrique et électronique de Paris (GeePs) %+ Systèmes et Applications des Technologies de l'Information et de l'Energie (SATIE) %A Cai, Caifang %A Lambert, Marc %A Rodet, Thomas %A Lesselier, Dominique %< avec comité de lecture %B 21th International Workshop on Electromagnetic Nondestructive Evaluation (ENDE 2016) %C Lisbonne, Portugal %Y H. G. Ramos and A. L. Ribeiro %3 Proc. 21th International Workshop on Electromagnetic Nondestructive Evaluation (ENDE 2016) %P Paper 21, 60-61 %8 2016-09-25 %D 2016 %K Eddy current modeling %K Non-destructive evaluation/testing %K Bayesian parameter inversion %K Benchmarks %K Experimental data analysis %Z Engineering Sciences [physics]/Signal and Image processing %Z Engineering Sciences [physics]/Electromagnetism %Z Mathematics [math]/Mathematical Physics [math-ph]Conference papers %X In eddy-current nondestructive testing (EC-NdT), parameter inversion is one main challenge. Here, one attempts to depict key advances and to put those in perspective with respect to challenges ahead in the real world of EC-NdT within a Bayesian framework. This work is based on the use of the well-known CIVA platform, to compute the variations of eddy-current impedance observed above work pieces that will serve us as synthetic data in the process, and also on the use of databases, those being properly constructed from CIVA, since we are keen to use meta-models (see later) that enable us to avoid in particular being burdened by possibly very high computational costs if left to call for CIVA solvers repeatedly. Data acquired in laboratory-controlled conditions, be they benchmarks or experimental data from partners, are also exploited, that is, they serve us as inputs of parameter inversion, in order to provide further means of analysis of pros and cons of the proposed solution. The work, which will be illustrated by ample numerical simulations and thorough controlled-laboratory data exploitation shown during the expected presentation, is as indicated being led within a Bayesian framework. %G English %L hal-01376774 %U https://centralesupelec.hal.science/hal-01376774 %~ UPMC %~ CNRS %~ UNIV-CERGY %~ UNIV-PSUD %~ ENS-CACHAN %~ CNAM %~ SUP_LSS %~ INSMI %~ SATIE %~ SUP_SIGNAUX %~ IFSTTAR %~ CENTRALESUPELEC %~ GEEPS %~ PIEM %~ TDS-MACS %~ UNIV-PARIS-SACLAY %~ UNIV-PSUD-SACLAY %~ CENTRALESUPELEC-SACLAY %~ ENS-CACHAN-SACLAY %~ UPMC_POLE_1 %~ SATIE-SIAME %~ SATIE-MOSS %~ UNIV-RENNES %~ SORBONNE-UNIVERSITE %~ SU-SCIENCES %~ CY-TECH-SE %~ ANR %~ FARMAN %~ ENS-PARIS-SACLAY %~ GS-ENGINEERING %~ GS-COMPUTER-SCIENCE %~ GS-PHYSIQUE %~ INSTITUT-SCIENCES-LUMIERE %~ ALLIANCE-SU %~ UNIV-EIFFEL %~ IFSTTAR-UNIVEIFFEL %~ TEST3-HALCNRS %~ HESAM-CNAM %~ HESAM