An iterative algorithm for eddy current NDE signal inversion
Date
2000
Authors
Liu, Guizhong
Major Professor
Advisor
Upda, Satish S.
Committee Member
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
A significant amount of research activity has been carried out in recent times for solving inverse problems in NDE using non-phenomenological approaches. Such applications typically involve the use of artificial neural networks, pattern recognition algorithms, etc. However, with the rapid advances in computational capabilities, phenomenological inverse models that simulate the forward problem and iteratively arrive at a solution are attracting considerable attention. This thesis presents an iterative algorithm for inverting eddy current NDE signals. The objective of this work is to estimate parameters characterizing the defect size and shape based on information contained in the measured eddy current signals. The approach uses the Finite Element Method (FEM) to model the physical process. The parameters of the forward model are updated using an iterative scheme that minimizes a cost function which is related to the error between the measured and FEM predicted signals. The cost function is minimized by updating the defect profile parameters. The conjugate gradient method is used to arrive at the minima in the multidimensional search space. Simulation results, obtained using an axisymmetric test geometry, demonstrate the effectiveness of the approach.
Series Number
Journal Issue
Is Version Of
Versions
Series
Academic or Administrative Unit
Type
thesis