Clutter Removal and Inversion of Eddy-Current Impedance Data (Postprint)

摘要

A wide variety of problems in computational electromagnetics have been successfully solved using a volume-integral approach along with conjugate- gradient methods. The volume integral algorithm is particularly well suited to the notion of model-based inversion of eddy-current impedance data in the arena of quantitative nondestructive evaluation (NDE). The first step in preparing data for inversion is to remove 'clutter,' which can originate in systematic measurement errors, or simply be a large background signal. In this paper we develop a systematic analytical technique to remove such clutter, which then allows the resulting impedance data to be fed into a nonlinear least-squares inversion algorithm, NLSE. The computational engine for the analysis is VIC-3D , a proprietary volume-integral code.