Deep recurrent neural networks with spectral-statistical fusion for industrial-grade steel alloy classification using femtosecond laser-ablation spark-induced breakdown spectroscopy
High-precision industrial steel alloy classification via fa-LA-SIBS and DeepRNN (GRU/LSTM/vanilla RNN) with spectral-statistical fusion. The DeepRNN framework models achieves 97.85%–99.00% average accuracy, with both macro- and micro-AUC >0.999.
Authors
- Xiaoyong He (ORCID: https://orcid.org/0000-0002-2379-2704)
- Zhenman Gao
- Xi Wu (ORCID: https://orcid.org/0009-0006-6457-8049)
- Tao Liang (ORCID: https://orcid.org/0000-0003-4627-8211)
- Kaiqiang Que
Institutions
- Dongguan University of Technology (CN)
- Nanning Normal University (CN)
Publication Details
- Journal
- Journal of Analytical Atomic Spectrometry
- Published
- 2025-10-28
- DOI
- https://doi.org/10.1039/d5ja00288e
- Primary Topic
- Laser-induced spectroscopy and plasma
- Type
- article
- Field-Weighted Citation Impact
- 0.00