Fundamental Feature Extraction Methods for the Analysis of Eddy Current Data (Preprint)

摘要

The objective of this paper is to explore features in eddy current data that are sensitive to defects in airframe structures while invariant to other noise factors commonly encountered in nondestructive evaluation (NDE). In particular, one goal is to detect and quantify corrosion-induced material loss in multi-layer aircraft structures. To investigate this problem, a series of eddy current studies were performed using an analytical model for varying total subsurface thickness loss (6%, 8%, and 10%), and percentage of the thickness loss occurring in the first or second layer (0, 25, 50, 75, and 100%). Results for the simulated studies with varying frequency are presented. A novel feature involving the first and second order derivatives of the real and imaginary parts of the impedance with respect to frequency is presented. Another goal is to detect and quantify subsurface cracks around fastener holes in structures. To investigate this problem, a series of studies are conducted using numerical models and then empirical data sets are analyzed to verify the viability of feature extraction techniques developed through the use of modeling. A feature sensitive to subsurface cracks around fastener holes is shown. Studies are conducted to show that this feature is invariant to irregular.