Fault Diagnosis and Early Warning Classification of Force-Sensing Optical Fibers: Physical Modeling, Anomaly Indexing, and Deep Learning Algorithms

DOI: https://doi.org/10.62517/jes.202602225

Author(s)

Xin Liu, Zhonglin Xu, Tao He, Hao Xiang, Jinhui Zhao, Yaoyi Jiao, Dai Hou, Junguo Hu, Chen Chen, Qian Wang, Bei Wang

Affiliation(s)

Hubei Siji Technology Co., Ltd., Wuhan, Hubei, China

Abstract

Force-sensing optical fibers have become an indispensable cornerstone in modern structural health monitoring, robotic tactile perception, and harsh-environment telemetry due to their immunity to electromagnetic interference and multiplexing capabilities. However, when deployed in complex, dynamic, and physically demanding environments, these waveguides are highly susceptible to mechanical degradation, static fatigue, and structural failures, which compromise data integrity and system safety. This paper presents an exhaustive, data-driven methodology for the real-time fault diagnosis and early warning classification of force-sensing optical fibers. We propose a hybrid algorithmic framework that bridges deterministic physical failure modeling with advanced machine learning. The methodology mathematically details the physical mechanisms of fiber buckling and crack propagation, utilizing wavelet packet decomposition for multi-resolution signal denoising. For early warning, we introduce a Synthetical Anomaly Index (SAI) that statistically aggregates temporal and spectral features to flag impending sensor failures before catastrophic signal loss. For precise fault classification, a hybrid Convolutional Neural Network and Bidirectional Long Short-Term Memory (CNN-BiLSTM) network is implemented, processing the optical data to categorize distinct failure modes. Experimental validations demonstrate that the SAI algorithm provides robust early warnings, while the CNN-BiLSTM model achieves a fault classification accuracy of 99.3% across eight distinct fault categories.

Keywords

Force-Sensing Optical Fiber; Fault Diagnosis; Early Warning Classification; Synthetical Anomaly Index; Machine Learning; CNN-BiLSTM; Structural Health Monitoring

References

[1]BAO X, WANG Y. Recent advancements in Rayleigh scattering-based distributed fiber sensors. Advances in Devices & Instrumentation, 2021, 2021: 8696571. [2]LAAROSSI I, ROLDÁN-VARONA P, POLO J A, et al. Sensing using light: a key area of sensors. Sensors, 2021, 21(19): 6562. [3]ZHANG Y, ZHANG X, WANG Y, et al. Fault diagnosis of rotating machinery based on parallel CNN-BiLSTM. Measurement, 2021, 182: 109723. [4]MUSTAFA S, SEKIYA H, MAEDA I, et al. Identification of external load information using distributed optical fiber sensors embedded in an existing road pavement. Optical Fiber Technology, 2021, 67: 102705. [5]LI Z, WANG Y, ZHANG P, et al. Underground power cable condition monitoring and risk assessment based on distributed perception and optical fiber sensing. IEEE Sensors Journal, 2024, 24(5): 1-10. [6]XIN C, ZHENG S, WANG X, et al. Physics-guided self-attention for metallic plate impact localization with FBG under low-sampling-rate constraints. IEEE Sensors Journal, 2024, 24(12): 1-10. [7]XU J, CHEN Y, LI H, et al. Deep learning for fault diagnosis in power transmission lines: current trends, limitations, and future directions. IEEE Access, 2025, 13: 1-15. [8]LIU Q, YU Y, HAN B S, et al. A vibration-based hybrid deep learning approach for anomaly detection of multi-joint industrial robots. IEEE Transactions on Industrial Electronics, 2024, 71: 1-10. [9]STEPHENS A F, BUSCH A, SALAMONSEN R F, et al. Rotary ventricular assist device control with a fiber Bragg grating pressure sensor. IEEE Transactions on Control Systems Technology, 2021, 29(3): 1009-1018. [10]ZENG Z, GHIASI A, NG C T, et al. Generalization of anomaly detection in bridge structures using a vibration-based Siamese convolutional neural network. Computer-Aided Civil and Infrastructure Engineering, 2025, 40(18): 1-18. [11]WANG Y, ZHANG X. Power grid fault diagnosis method based on CNN// 2021 International Conference on Power Energy Systems. IEEE, 2021: 1-5. [12]CONG T, TAN R, OTTEWILL J R, et al. Anomaly detection and mode identification in multimode processes using the field Kalman filter. IEEE Transactions on Control Systems Technology, 2021, 29(5): 2192-2205. [13]DA SILVA P M, MENDES J P, COELHO L C C, et al. Real-time monitoring of cement paste carbonation with in situ optical fiber sensors . Sensors, 2023, 11(8): 449. [14]TSAI W C, HONG C M, TU C S, et al. A review of modern wind power generation forecasting technologies. Sustainability, 2023, 15(14): 10757. [15]LIU J, ZHANG X, WANG Y, et al. Deep learning methodology employing Bayesian and active learning strategies for anomaly detection. Computational Materials Science, 2022, 200: 110800.