A K-SVD-Guided BiLSTM Network for Intelligent Classification of Rail Acoustic Emission Signals

Shuzhi Song; Yifei Chen; Zhengyu Chen; Xuemei Guan

doi:10.1007/978-981-95-8232-7_22

A K-SVD-Guided BiLSTM Network for Intelligent Classification of Rail Acoustic Emission Signals

Shuzhi Song
, Yifei Chen
, Zhengyu Chen
, Xuemei Guan^*

^*Corresponding author for this work

School of Astronautics

Northeast Forestry University

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

In order to enhance the recognition accuracy of rail damage signals and address data imbalance in acoustic emission (AE) classification, this paper proposes a k-singular value decomposition (K-SVD) guided bidirectional long short-term memory (BiLSTM) network for intelligent classification of rail AE signals. A multi-layer wavelet packet transform with a kurtosis–energy metric is applied for effective noise suppression. The improved K-SVD then learns shared and stage-specific atoms to extract discriminative sparse coefficients, which are fed into a BiLSTM network optimized by a combined cross-entropy and focal loss (BiLSTM-CEFL). Experimental studies show that the proposed method achieves high reconstruction fidelity and superior classification performance, with the accuracy of 93.7% and AUC of 0.9619. The proposed method effectively distinguishes AE signals from different stages under the unbalanced dataset and interference, providing a robust and interpretable solution for rail structural.

Original language	English
Title of host publication	Proceedings of the 4th International Conference on Sensing, Measurement, Communication and Internet of Things Technologies
Editors	Zhenyu Zhao, Peiquan Jin, Mingchuan Zhang
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	228-238
Number of pages	11
ISBN (Print)	9789819582310
DOIs	https://doi.org/10.1007/978-981-95-8232-7_22
State	Published - 2026
Event	4th International Conference on Sensing, Measurement, Communication and Internet of Things Technologies, SMC-IoT 2025 - Luoyang, China Duration: 28 Nov 2025 → 30 Nov 2025

Publication series

Name	Lecture Notes in Electrical Engineering
Volume	1597 LNEE
ISSN (Print)	1876-1100
ISSN (Electronic)	1876-1119

Conference

Conference	4th International Conference on Sensing, Measurement, Communication and Internet of Things Technologies, SMC-IoT 2025
Country/Territory	China
City	Luoyang
Period	28/11/25 → 30/11/25

Keywords

Acoustic Emission
BiLSTM
Deep Learning
Intelligent Classification
K-SVD

Access to Document

10.1007/978-981-95-8232-7_22

Cite this

Song, S., Chen, Y., Chen, Z., & Guan, X. (2026). A K-SVD-Guided BiLSTM Network for Intelligent Classification of Rail Acoustic Emission Signals. In Z. Zhao, P. Jin, & M. Zhang (Eds.), Proceedings of the 4th International Conference on Sensing, Measurement, Communication and Internet of Things Technologies (pp. 228-238). (Lecture Notes in Electrical Engineering; Vol. 1597 LNEE). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-981-95-8232-7_22

Song, Shuzhi ; Chen, Yifei ; Chen, Zhengyu et al. / A K-SVD-Guided BiLSTM Network for Intelligent Classification of Rail Acoustic Emission Signals. Proceedings of the 4th International Conference on Sensing, Measurement, Communication and Internet of Things Technologies. editor / Zhenyu Zhao ; Peiquan Jin ; Mingchuan Zhang. Springer Science and Business Media Deutschland GmbH, 2026. pp. 228-238 (Lecture Notes in Electrical Engineering).

@inproceedings{c5995eaae33043f98c775544633bd1a0,

title = "A K-SVD-Guided BiLSTM Network for Intelligent Classification of Rail Acoustic Emission Signals",

abstract = "In order to enhance the recognition accuracy of rail damage signals and address data imbalance in acoustic emission (AE) classification, this paper proposes a k-singular value decomposition (K-SVD) guided bidirectional long short-term memory (BiLSTM) network for intelligent classification of rail AE signals. A multi-layer wavelet packet transform with a kurtosis{\textendash}energy metric is applied for effective noise suppression. The improved K-SVD then learns shared and stage-specific atoms to extract discriminative sparse coefficients, which are fed into a BiLSTM network optimized by a combined cross-entropy and focal loss (BiLSTM-CEFL). Experimental studies show that the proposed method achieves high reconstruction fidelity and superior classification performance, with the accuracy of 93.7\% and AUC of 0.9619. The proposed method effectively distinguishes AE signals from different stages under the unbalanced dataset and interference, providing a robust and interpretable solution for rail structural.",

keywords = "Acoustic Emission, BiLSTM, Deep Learning, Intelligent Classification, K-SVD",

author = "Shuzhi Song and Yifei Chen and Zhengyu Chen and Xuemei Guan",

note = "Publisher Copyright: {\textcopyright} The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026.; 4th International Conference on Sensing, Measurement, Communication and Internet of Things Technologies, SMC-IoT 2025 ; Conference date: 28-11-2025 Through 30-11-2025",

year = "2026",

doi = "10.1007/978-981-95-8232-7\_22",

language = "英语",

isbn = "9789819582310",

series = "Lecture Notes in Electrical Engineering",

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "228--238",

editor = "Zhenyu Zhao and Peiquan Jin and Mingchuan Zhang",

booktitle = "Proceedings of the 4th International Conference on Sensing, Measurement, Communication and Internet of Things Technologies",

address = "德国",

}

Song, S, Chen, Y, Chen, Z & Guan, X 2026, A K-SVD-Guided BiLSTM Network for Intelligent Classification of Rail Acoustic Emission Signals. in Z Zhao, P Jin & M Zhang (eds), Proceedings of the 4th International Conference on Sensing, Measurement, Communication and Internet of Things Technologies. Lecture Notes in Electrical Engineering, vol. 1597 LNEE, Springer Science and Business Media Deutschland GmbH, pp. 228-238, 4th International Conference on Sensing, Measurement, Communication and Internet of Things Technologies, SMC-IoT 2025, Luoyang, China, 28/11/25. https://doi.org/10.1007/978-981-95-8232-7_22

A K-SVD-Guided BiLSTM Network for Intelligent Classification of Rail Acoustic Emission Signals. / Song, Shuzhi; Chen, Yifei; Chen, Zhengyu et al.
Proceedings of the 4th International Conference on Sensing, Measurement, Communication and Internet of Things Technologies. ed. / Zhenyu Zhao; Peiquan Jin; Mingchuan Zhang. Springer Science and Business Media Deutschland GmbH, 2026. p. 228-238 (Lecture Notes in Electrical Engineering; Vol. 1597 LNEE).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - A K-SVD-Guided BiLSTM Network for Intelligent Classification of Rail Acoustic Emission Signals

AU - Song, Shuzhi

AU - Chen, Yifei

AU - Chen, Zhengyu

AU - Guan, Xuemei

N1 - Publisher Copyright: © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026.

PY - 2026

Y1 - 2026

N2 - In order to enhance the recognition accuracy of rail damage signals and address data imbalance in acoustic emission (AE) classification, this paper proposes a k-singular value decomposition (K-SVD) guided bidirectional long short-term memory (BiLSTM) network for intelligent classification of rail AE signals. A multi-layer wavelet packet transform with a kurtosis–energy metric is applied for effective noise suppression. The improved K-SVD then learns shared and stage-specific atoms to extract discriminative sparse coefficients, which are fed into a BiLSTM network optimized by a combined cross-entropy and focal loss (BiLSTM-CEFL). Experimental studies show that the proposed method achieves high reconstruction fidelity and superior classification performance, with the accuracy of 93.7% and AUC of 0.9619. The proposed method effectively distinguishes AE signals from different stages under the unbalanced dataset and interference, providing a robust and interpretable solution for rail structural.

AB - In order to enhance the recognition accuracy of rail damage signals and address data imbalance in acoustic emission (AE) classification, this paper proposes a k-singular value decomposition (K-SVD) guided bidirectional long short-term memory (BiLSTM) network for intelligent classification of rail AE signals. A multi-layer wavelet packet transform with a kurtosis–energy metric is applied for effective noise suppression. The improved K-SVD then learns shared and stage-specific atoms to extract discriminative sparse coefficients, which are fed into a BiLSTM network optimized by a combined cross-entropy and focal loss (BiLSTM-CEFL). Experimental studies show that the proposed method achieves high reconstruction fidelity and superior classification performance, with the accuracy of 93.7% and AUC of 0.9619. The proposed method effectively distinguishes AE signals from different stages under the unbalanced dataset and interference, providing a robust and interpretable solution for rail structural.

KW - Acoustic Emission

KW - BiLSTM

KW - Deep Learning

KW - Intelligent Classification

KW - K-SVD

UR - https://www.scopus.com/pages/publications/105042542354

U2 - 10.1007/978-981-95-8232-7_22

DO - 10.1007/978-981-95-8232-7_22

M3 - 会议稿件

AN - SCOPUS:105042542354

SN - 9789819582310

T3 - Lecture Notes in Electrical Engineering

SP - 228

EP - 238

BT - Proceedings of the 4th International Conference on Sensing, Measurement, Communication and Internet of Things Technologies

A2 - Zhao, Zhenyu

A2 - Jin, Peiquan

A2 - Zhang, Mingchuan

PB - Springer Science and Business Media Deutschland GmbH

T2 - 4th International Conference on Sensing, Measurement, Communication and Internet of Things Technologies, SMC-IoT 2025

Y2 - 28 November 2025 through 30 November 2025

ER -

Song S, Chen Y, Chen Z, Guan X. A K-SVD-Guided BiLSTM Network for Intelligent Classification of Rail Acoustic Emission Signals. In Zhao Z, Jin P, Zhang M, editors, Proceedings of the 4th International Conference on Sensing, Measurement, Communication and Internet of Things Technologies. Springer Science and Business Media Deutschland GmbH. 2026. p. 228-238. (Lecture Notes in Electrical Engineering). doi: 10.1007/978-981-95-8232-7_22

A K-SVD-Guided BiLSTM Network for Intelligent Classification of Rail Acoustic Emission Signals

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this