Comparison of efficient global optimization and particle swarm optimization for eddy current non-destructive evaluation

This paper presents a comparison of the Efficient Global Optimization (EGO) and Particle Swarm Optimization (PSO) algorithms in the context of defect characterization in Eddy current non-destructive evaluation (EC-NDE). An axisymmetric internal groove is applied to an electric conducting tube; the forward model linking measurements to electromagnetic properties of the tube is solved using the finite element method FEM. Both PSO and EGO techniques are used to estimate the geometric parameters of the groove. The main result of the work is to show the efficiency of the use of surrogate models based on Design and Analysis of Computer Experiments (DACE) in Eddy current inversion in terms of reduction of the computing time.