Crack Estimation Using Adaptive Inversion Database with Multi Output Support Vector Machine for Eddy Current Testing

This paper considers the inverse problem of estimating the size of an unknown defect from the measurement of the impedance variations, By the novel combination of the multi output support vector machines (MO-SVM), and finite element modeling for eddy current flaw Characterization of a defect size in a conductive nonmagnetic plate.The finite element method (FEM) is used to create the database, In this context, the cylindrical sensor have been considered to validate a COMSOL-Multi physics 3D-resolution using a 3D electromagnetic formulation , A good agreement is obtained between the numerical results and the experimental ones. The database required to train the MO-SVM , Several method are used to find the parameters of MOSVM. the Particle Swarm Optimization (PSO) ,Genetic Algorithm GA are proposed.