Optimal database combining with Multi Output Support Vector Machine for Eddy Current Testing

M. Chelabi , Tarik Hacib , Z. Belli , Mohamed Mekideche , Yann Le Bihan (1)

1 LGEP - Laboratoire de génie électrique de Paris

This paper provides a new methodology for the characterization of defect size in a conductive nonmagnetic plate from the measurement of the impedance variations. The methodology is based on Finite Element Method (FEM) combined with the Multi Output Support Vector Machines (MO-SVM). The MO-SVM is a statistical learning method that has good generalization capability and learning performance. FEM is used to create the adaptive database required to train the MO-SVM and the Cross Validation (CV) is used to find the parameters of MO-SVM model. The results show the applicability of MO-SVM to solve eddy current inverse problems instead of using traditional iterative inversion methods which can be very time-consuming. With the experimental results we demonstrate the accuracy which can be provided by the MO-SVM technique.

Domains: Engineering Sciences [physics]

M. Chelabi, Tarik Hacib, Z. Belli, Mohamed Mekideche, Yann Le Bihan. Optimal database combining with Multi Output Support Vector Machine for Eddy Current Testing. EVER 2014, Mar 2014, Monaco, Monaco. pp.1 - 4, ⟨10.1109/EVER.2014.6844142⟩. ⟨hal-01099330⟩