Dynamic classification of bearing steel based on pulsed eddy current
Measurement Science and Technology · 17.04.2026 · 37 (16) · 166103
DOI: 10.1088/1361-6501/ae5df3 ↗Kurz gefasst
Fachartikel in der Zeitschrift Measurement Science and Technology, online erschienen am 20.04.2026. Verfasst von Miao Qian und 3 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 34 Literaturangaben. Gefördert von: Key R&D projects of Science and Technology Department of Zhejiang Province.
Abstract (Originalsprache)
Bearings, as the core components of mechanical equipment, have performance that directly depends on the correct selection of raw material bearing steel. This paper proposes a dynamic non-destructive testing method based on differential pulsed eddy current, combined with a dual-branch feature fusion neural network, achieving classification of known classes and recognition of unknown classes for moving bearing steel. Firstly, a dynamic dual-probe differential acquisition system was designed to stably obtain eddy current response signals under the axial motion of the steel. Secondly, a dual-branch convolutional neural network was developed to fuse spatio-temporal and statistical features of the signals. Adaptive feature fusion was achieved through an attention mechanism, and a dual-threshold open-set recognition strategy combined with a multiple-sampling decision fusion mechanism improved the classification accuracy for both known and unknown classes. Finally, experiments were conducted on various types of steel. The experimental results demonstrate that in open-set tests, the proposed method achieved an average classification accuracy exceeding 96.7% for 5 known steel types (e.g., GCr15, 100Cr6), and an average unknown class recognition accuracy exceeding 93.3% for unknown steel types (e.g., 9Cr18). Utilizing the three-time sampling strategy, an overall misclassification rate nearly 0% was observed within the test dataset, while the system maintained robustness across varying speeds. The proposed system provides a practical solution for the online real-time sorting of bearing steel raw materials.
Förderung
- Key R&D projects of Science and Technology Department of Zhejiang Province
Zitieren
APA
Qian, M., Yu, X., Xiang, Z., & Zhang, J. (2026). Dynamic classification of bearing steel based on pulsed eddy current. Measurement Science and Technology, 37(16), 166103. https://doi.org/10.1088/1361-6501/ae5df3
Vancouver
Qian M, Yu X, Xiang Z, Zhang J. Dynamic classification of bearing steel based on pulsed eddy current. Measurement Science and Technology. 2026;37(16):166103. doi:10.1088/1361-6501/ae5df3
BibTeX
@article{Qian2026,
author = {Qian, Miao and Yu, Xiaoxiao and Xiang, Zhong and Zhang, Jianxin},
title = {Dynamic classification of bearing steel based on pulsed eddy current},
journal = {Measurement Science and Technology},
year = {2026},
volume = {37},
number = {16},
pages = {166103},
doi = {10.1088/1361-6501/ae5df3}
}
EUVIMED