Real Time Pulsed Eddy Current Detection of Cracks in F/A-18 Inner Wing Spar Using Discriminant Separation of Modified Principal Components Analysis Scores

Abstract

The aluminum inner wing spars of F/A-18 Hornet aircraft may undergo stress corrosion cracking along the spar between the fasteners that secure carbon-fiber/epoxy composite skin to the wing. Inspection of the spar through the wing skin is required to avoid wing disassembly. The thickness of the skin varies between 9 and 21 mm (0.36-0.83 in) and fasteners may be either titanium or ferrous. Pulsed eddy currents (PECs), generated by a probe centered over the fastener, demonstrate the capability of detecting simulated cracks within spars with the wing skin present. Comparison of signals from separate pick-up coils, mounted to either side of the excitation coil, is used to detect differences in induced eddy current fields, which arise in the presence of cracks. To overcome variability in PEC signal response due to variation in: 1) skin thickness; 2) fastener material and size; and 3) centering over fasteners, a large calibration data set is acquired. Multidimensional scores from a modified principal components analysis (PCA) of the data are reduced to 1-D using a discriminant analysis method. Under inspection conditions, calibration PCA scores combined with discriminant analysis permit rapid real time go/no-go PEC detection of cracks in F/A-18 inner wing spar.

Pulsed Eddy Current Non-destructive Testing and Evaluation: A Review

“…Other examples of work where PCA has been used are introduced in Refs. [14], [28], [48], [52], [55][56][57][58][59][60].…”

Enhancing pulsed eddy current for inspection of P-3 Orion lap-joint structures

“…This covariance matrix is then used to calculate the MD for all experimental data signals. The MD is then compared to a decision threshold in order to determine if the fastener hole has a notch present or not [1,7]. Figure 5 shows a 3-D view of MPCA scores and MD for data collected using the P-3 Orion sample.…”

Pulsed Eddy Current Detection of Cracks in Multilayer Aluminum Lap Joints

“…A relatively small number of vectors reproduce the original signal, typically 2-3 [18], [17], thereby reducing all the variation among the data to a manageable number of scores. The full derivation of modified PCA, which uses a least-squares method of finding the best set of basis vectors to represent the original signal without average signal subtraction, is described in detail elsewhere [16].…”

“…A number of techniques have been proposed to extract this information including Independent Component Analysis (ICA) [11], Principal Components Analysis (PCA) [12], [13], PCA combined with wavelet analysis [14] and Smooth Nonnegative Matrix Factorization [15], to mention a few. Recently, a modified form of PCA [16] has been used to decompose differential probe signals and successfully detect simulated stress corrosion cracking in the inner wing spars of F/A-18 aircraft without skin removal [17] and this is the approach that is used in this work. The present work supports recent finite element modeling (FEM) results [18] that demonstrated the potential of PEC as a technique for second layer crack detection in aluminum lap joints.…”

“…Discriminant analysis has been used to create separation between scores from cracked and un-cracked fastener sites using PEC technique [17], where scores from cracked sites congregated to one side of the cluster of scores from un-cracked sites. However, the FEM [18] showed that, in the current application, crack orientation and size affects score position, i.e.…”