Sparse grids for eddy-current non-destructive testing

Sándor Bilicz (1) , Caifang Cai (2) , Marc Lambert (3)

1 BME - Budapest University of Technology and Economics [Budapest]
2 L2S - Laboratoire des signaux et systèmes
3 GeePs - Laboratoire Génie électrique et électronique de Paris

Sparse grids make possible the interpolation-based approximation of the output of computer experiments, even for a high number of independent input parameters. In this contribution, the sparse grid interpolation technique is shown to apply well to eddy-current non-destructive testing and simulations thereof. Once such interpolation is obtained –being computationally cheap–, the use of simple optimisation schemes for the inversion becomes numerically efficient. A common pattern-search algorithm illustrates this concept.

Keywords: non-destructive testing; sparse grids; surrogate modeling; parameter inversion; Markov Chain Monte-Carlo

Domains: Signal and Image processing; Electromagnetism; Mathematical Physics [math-ph]

Sándor Bilicz, Caifang Cai, Marc Lambert. Sparse grids for eddy-current non-destructive testing. 17th International Symposium on Applied Electromagnetics and Mechanics (ISEM2015), Sep 2015, Awaji City, Japan. Paper OS1-4 (2 p.). ⟨hal-01222776⟩