Eddy Current Classification of Sheet Ferromagnetic Materials Based on Material Property Profiles.

This paper proposes a novel strategy for the fine-grade classification of sheet ferromagnetic materials with similar electromagnetic properties. The strategy integrates three original components: suppression of the effects of uncontrollable factors on the probe signal, reconstruction of material property profiles, and classification in a reduced-dimensional latent space. Together, these components constitute an innovative framework for effectively identifying sheet samples using the eddy current method. The strategy can be summarized as follows: interference-suppressed eddy current probe signal → GAN latent space → distance-to-class classifier → material grade. This part of the study focuses on generalizing, extending, and adapting the method previously developed by the authors for acquiring signals directly from a Taguchi-optimized probe without prior signal processing. The method is applied to the indirect measurement of material property profiles, which constitutes the first stage and a key prerequisite for implementing the proposed sample-identification concept. The results of the numerical experiments demonstrate that the optimally designed probe provides a high signal-to-noise ratio and an acceptable reduction in output-signal variations caused by uncontrollable factors.