ECT inversion using a knowledge-based forward solver

In this paper, a new approach was proposed for crack reconstruction based on ECT technique and optimization theory. A newly developed fast forward solver was applied for predicting the impedance signals during the inversion procedure. Comparing with the conventional approaches, the scanning signal due to an arbitrary shaped crack can be calculated in a significantly reduced CPU time but with higher accuracy. The conjugate gradient algorithm was employed in the optimization of the crack parameters, with the gradient calculated numerically from the electric field in the crack region. The scheme was implemented for the shape reconstruction of EDM cracks from the measured impedance data having in mind possible applications in the inspection of steam generator tubing. Several cracks embedded in conducting tube or plate were reconstructed from the observed impedance data by concerning the benchmark models proposed by T. Takagi et al. Good reconstruction results were obtained within a short CPU time. Consequently, the validity of the new approach was verified.