Some advances in numerical analysis techniques for quantitative electromagnetic nondestructive evaluation

Quantitative non-destructive evaluation of wall thinning defect in double-layer pipe of nuclear power plants using pulsed ECT method

“…To deal with three-dimensional wall thinning (volume defect) geometry, the conventional fast simulator for crack like defects [15][16][17] has been upgraded. In addition to the improved database approach, the fast PECT forward solver is also upgraded with the aid of an interpolation strategy.…”

“…Recently, a high efficiency 3D numerical tool (a fast forward solver) to predict PECT signals has been developed [14] by authors. This 3D code is based on a frequency domain computation scheme and employs a database strategy which is similar to that developed for the fast simulation of single frequency ECT signals [15][16][17]. The 3D code gives a good basis for the quantitative wall thinning analysis based on inversion techniques.…”

Reconstruction of cracks in a carbon fiber-reinforced polymer laminate plate from signals of eddy current testing

“…For reconstruction of crack from ECT signals with the use of an optimization method, a lot of forward signal simulations have to be conducted for both the deterministic and stochastic inversion schemes. 18–20 Therefore, a fast forward simulation scheme of high accuracy for the anisotropic material is very important for sizing crack in CFRP material. In the following, the fast ECT forward solver updated by the authors to cope with the anisotropic CFRP material problem based on the ECT of isotropic metallic materials 21 is introduced.…”

Three‐dimensional defect inversion from magnetic flux leakage signals using iterative neural network

“…Considering the complex numerical computation of solving the ill-posed inverse MFL problem [5,6], most defect inversion methods nowadays are based on two-dimensional (2D) other than 3D defect profile. There are generally two kinds of methods to solve the inverse MFL problem, including the non-model-based direct method and the model-based iterative method [7]. The non-model-based direct method takes use of neural networks or other tools to establish a mapping relationship between MFL signals and corresponding defect profiles [8,9].…”