A model for inverting eddy-current data and reconstructing three-dimensional flaws embedded in anisotropic advanced composite materials

The authors present a three-dimensional linearized model for inverting eddy-current data to reconstruct flaws in advanced anisotropic composite materials, such as graphite epoxy. This problem is significant for enhancing the detection capability for the variety of failure modes in composites, such as delamination, fiber breakage, flaws, etc. The inversion scheme takes the form of an integral equation, which then is discretized by means of the method of moments. The measured data are then inverted by means of the conjugate gradient algorithm. Examples are shown in which a linear classifier is used to improve convergence of the conjugate gradient algorithm. As an example, a 0.11-in.-thick woven graphite epoxy material was simulated for the inversion algorithm.<
>