A learning-by-examples approach for non-destructive localization and characterization of defects through eddy current measurements

This work presents an innovative inversion method for non-destructive localization and characterization of defects based on a Learning-by-Example (LBE) approach. The qualitative imaging of the electromagnetic anomalies affecting a conductive structure is based on the exploitation of Eddy Current Testing (ECT) measurements through Support Vector Regression (SVR), enabling an evaluation of the structure under test almost in real-time. Some preliminary numerical results are shown in order to validate the effectiveness of the proposed method.