TY  - RPRT
ID  - elediasc12714
A1  - Salucci, M.
A1  - Anselmi, N.
A1  - Oliveri, G.
A1  - Massa, A.
KW  - Eddy current testing
KW  -  inverse scattering
KW  -  nondestructive testing and evaluation
KW  -  statistical learning
KW  -  learning-by-examples
KW  -  support vector regression
PB  - University of Trento
AV  - public
UR  - http://www.eledia.org/students-reports/714/
Y1  - 2016///
TI  - NDT-NDE Crack Characterization Through a Learning-by-Examples Approach
N2  - This document deals with the characterization of a single narrow crack in a planar conductive structure starting from eddy current testing (ECT) measurements. More precisely, the inversion problem at hand is formulated within the so-called learning-by-examples (LBE) paradigm, by considering the problem of estimating the dimensions of the defect as a regression one. Accordingly, a set of known input-output pairs is generated during an off-line phase and is given as input to a Support Vector Regressor (SVR) prediction model in order to train it on the relationship between defect and corresponding ECT data. Some numerical results are shown in order to verify the effectiveness, as well as the limits, of the proposed LBE technique when dealing with the presence of noise on testing data during the on-line inversion phase.
M1  - technical_report
ER  -