Application of projection methods of multivariate data analysis in eddy current testing of materials

Abstract

The paper considers the applicability of projection methods of multivariate data analysis to discriminate between the factors that simultaneously affect the results of multi-frequency eddy current testing of nonmagnetic metals and alloys. Measurements were carried out for copper, magnesium, aluminum alloy and bronze specimens with different electrical conductivity equal to 57, 22, 16 and 7.5 S/m, respectively. The measured probe impedance changes were used to plot hodographs within the frequency range from 100 Hz to 6.4 kHz. The gap width between an attachable parametric probe and the specimen surface was specified using dielectric spacers within the range from 0 to 1 mm. The principal component analysis applied to experimental hodographs allowed us to safely discriminate between the influence of such factors as electrical conductivity of the material and gap width. The proposed approach to discriminating between individual factors that strongly affect eddy current measurement results is an enhancement in eddy current testing of materials.