Volume 10 • Issue 1 • PP: 26–34 • 2026
TA-FaultNet: A Temporal Attention Framework with Bidirectional LSTM for Multi-Class Fault Detection and Health Monitoring in Industrial Wireless Sensor Networks
View PDF Abstract
Industrial wireless sensor networks are central to the continuous monitoring of critical plant equipment, yet reliable identification of multiple concurrent fault modes from heterogeneous multivariate sensor streams remains an unsolved operational challenge. Physical failure mechanisms—pump cavitation, valve blockage, gradual sensor drift—and wireless channel disturbances each imprint distinct but overlapping temporal signatures that render classical thresholdand rule-based detectors inadequate for automated maintenance dispatch. This paper presents TA-FaultNet, a neural architecture designed specifically for the multi-class fault identification problem in industrial sensor deployments. The network couples a two-stage stacked bidirectional recurrent encoder with a parallel multi-head self-attention module and a compact temporal convolutional block, enabling simultaneous capture of long-range process dynamics and fine-grained fault-onset localisation from raw sensor windows. TA-FaultNet is evaluated on the publicly available Skoltech Anomaly Benchmark under five operational classes and assessed through a comprehensive battery of experiments including baseline comparisons, systematic component ablation, cross-experiment generalisation, andprogressive noise-injection testing. The proposed architecture decisively outperforms eight competing methods spanning classical anomaly detectors, standalone recurrent and convolutional networks, and the Transformer, while remaining lightweight enough for edge gateway deployment. Attention weight visualisations expose fault-specific temporal activation patterns, providing maintenance engineers with interpretable diagnostic evidence beyond bare classification labels.
Keywords
References
[1] A. S. Balobaid, S. B. Ahamed, S. Shamsudheen, and S. Balamurugan, “Neural network clustering and swarm intelligence-based routing protocol for wireless sensor networks,” Wireless Communications and Mobile Computing, vol. 2023, p. 4758852, 2023, doi: 10.1155/2023/4758852.
[2] O. A. Khashan, N. M. Khafajah, W. Alomoush, and M. Alshinwan, “Innovative energy-efficient proxy reencryption for secure data exchange in wireless sensor networks,” IEEE Access, vol. 12, pp. 23 290–23 304, 2024, doi: 10.1109/ACCESS.2024.3360488.
[3] H. Ruan, B. Dorneanu, H. Arellano-Garcia, P. Xiao, and L. Zhang, “Deep learning-based fault prediction in wireless sensor network embedded cyberphysical systems for industrial processes,” IEEE Access, vol. 10, pp. 10 867–10 879, 2022, doi: 10.1109/ACCESS. 2022.3144333.
[4] Y. Liu, S. Garg, J. Nie, Y. Zhang, Z. Xiong, J. Kang, and M. S. Hossain, “Deep anomaly detection for timeseries data in industrial IoT: A communication-efficient on-device federated learning approach,” IEEE Internet of Things Journal, vol. 8, no. 8, pp. 6348–6358, 2021, doi: 10.1109/JIOT.2020.3014586.
[5] X. Wen and W. Li, “Time series prediction based on LSTM-Attention-LSTM model,” IEEE Access, vol. 11, pp. 48 322–48 331, 2023, doi: 10.1109/ACCESS. 2023.3276628.
[6] I. D. Katser and V. O. Kozitsin, “Skoltech anomaly benchmark (SKAB),” 2020, doi: 10.34740/KAGGLE/ DSV/1693952.
[7] A. Bagwari, J. Logeshwaran, K. Usha, R. Kannadasan, M. H. Alsharif, P. Uthansakul, and M. Uthansakul, “An enhanced energy optimization model for industrial wireless sensor networks using machine learning,” IEEE Access, vol. 11, p. 3311854, 2023, doi: 10.1109/ACCESS. 2023.3311854.
[8] M. Aljawarneh, R. Hamdaoui, A. Zouinkhi, S. Alangari, and M. N. Abdelkrim, “Energy optimization for wireless sensor network using minimum redundancy maximum relevance feature selection and classification techniques,” PeerJ Computer Science, vol. 10, p. e1997, 2024, doi: 10.7717/peerj-cs.1997.
[9] K. Rashid, Y. Saeed, A. Ali, F. Jamil, R. Alkanhel, and A. Muthanna, “An adaptive real-time malicious node detection framework using machine learning in vehicular ad-hoc networks (VANETs),” Sensors, vol. 23, no. 5, p. 2594, 2023, doi: 10.3390/s23052594.
[10] N. S. Albalawi, Y. Alzahrani, N. Alsalmi, Y. Patidar, and M. Tolani, “Energy-efficient priority encoding strategies using machine learning based hybrid MAC protocol for wireless sensor networks,” Scientific Reports, vol. 15, p. 45054, 2025, doi: 10.1038/s41598-025-31752-1.
[11] D. Godfrey, B. Suh, B.-H. Lim, K.-C. Lee, and K.-I. Kim, “An energy-efficient routing protocol with reinforcement learning in software-defined wireless sensor networks,” Sensors, vol. 23, no. 20, p. 8435, 2023, doi: 10.3390/s23208435.
[12] X. Zhong, Y. Liang, and Y. Li, “Energy-efficient and robust QoS control for wireless sensor networks using the extended Gur game,” Sensors, vol. 25, no. 3, p. 730, 2025, doi: 10.3390/s25030730.
[13] N. V. Chawla, K. W. Bowyer, L. O. Hall, and W. P. Kegelmeyer, “SMOTE: Synthetic minority oversampling technique,” in Journal of Artificial Intelligence Research, vol. 16, 2002, pp. 321–357, doi: 10.1613/jair.953.
[14] S. Hochreiter and J. Schmidhuber, “Long short-term memory,” Neural Computation, vol. 9, no. 8, pp. 1735– 1780, 1997, doi: 10.1162/neco.1997.9.8.1735.
[15] F. Pedregosa, G. Varoquaux, A. Gramfort, V. Michel, B. Thirion, O. Grisel, M. Blondel, P. Prettenhofer, R.Weiss, V. Dubourg, J. Vanderplas, A. Passos, D. Cournapeau, M. Brucher, M. Perrot, and E. Duchesnay, “Scikit-learn: Machine learning in Python,” Journal of Machine Learning Research, vol. 12, pp. 2825–2830, 2011.
[16] V. K. Pandey, S. Prakash, T. K. Gupta, P. Sinha, T. Yang, R. S. Rathore, L. Wang, S. Tahir, and S. T. Bakhsh, “Enhancing intrusion detection in wireless sensor networks using a Tabu search based optimized random forest,” Scientific Reports, vol. 15, pp. 1–19, 2025, doi: 10.1038/s41598-025-03498-3.
Cite This Article
Choose your preferred format