AUTHOR=Kong Qingxiao , Pan Shuwei , Lin Lilong , Li Xinfeng TITLE=Classification, identification, and quantitative study of defects in aluminum plates using pulsed alternating current field measurement combined with principal component analysis JOURNAL=Frontiers in Materials VOLUME=Volume 12 - 2025 YEAR=2025 URL=https://www.frontiersin.org/journals/materials/articles/10.3389/fmats.2025.1569055 DOI=10.3389/fmats.2025.1569055 ISSN=2296-8016 ABSTRACT=IntroductionThis study investigates an approach for defect characterization in non-ferromagnetic materials by combining Pulsed Alternating Current Field Measurement (PACFM) with Principal Component Analysis (PCA). The research demonstrates how this integrated method can effectively classify and quantify both surface and subsurface defects through signal processing of PACFM data.MethodsThe PACFM technique was utilized to acquire defect response signals from non-ferromagnetic specimens. Subsequently, PCA was implemented to decompose the multidimensional PACFM datasets into principal components, with each component preserving the most diagnostically significant information. In this analytical framework, the classification of defects was determined by the sign of the mapped value w2 in the PCA eigenvector direction, while the magnitude of w2 exhibited a correlation with subsurface defect burial depths.ResultsThe integrated PACFM-PCA approach successfully discriminated between surface and subsurface defects. The polarity of the principal component w2 served as a reliable feature for defect classification, with positive values consistently corresponding to subsurface defects and negative values indicating surface defects. Furthermore, a robust quadratic relationship correlation was established between the eigenspace coordinates of subsurface defect signals and their respective burial depths, enabling accurate quantitative assessment of burial depth.DiscussionThe integration of PACFM with PCA provides a robust framework for defect analysis in non-ferromagnetic materials. This synergistic approach demonstrates significant capability in extracting and quantifying defect signatures from complex response signals, highlighting its considerable potential for non-destructive testing (NDT) applications. Future work could explore its adaptability to more intricate defect geometries.