Graph-Based Temporal Data Analytics for Predictive Infrastructure Monitoring
Keywords:
Predictive Maintenance, Graph Neural Networks, Temporal Graphs, Infrastructure Mon-itoring, Spatio-Temporal Analytics, Anomaly DetectionAbstract
Infrastructure systems such as bridges, railways, and pipelines generate complex time-dependent data streams that require advanced analytics to detect early signs of deterioration or failure. This paper proposes a graph-based temporal analytics framework that models infrastructure sensor data as dynamic graphs, allowing spatio-temporal relationships and structural dependencies to be encoded effectively. Leveraging graph neural networks (GNNs) and time-series learning, the approach enables predictive monitoring and early warning systems that outperform traditional flat data models. Evaluations demonstrate significant improvements in anomaly detection, failure prediction accuracy, and interpretability. This research outlines a scalable, robust solution for critical infrastructure health monitoring.
References
Wu, Y., Zhang, T., Lin, J., & Wang, F. (2023). Spatio-temporal graph learning for traffic flow prediction. IEEE Transactions on Neural Networks and Learning Systems, 34(5), 2341–2354.
Zhou, H., Wei, X., & Ma, Y. (2022). Dynamic graph convolutional networks for bridge structural health monitoring using vibration data. Structural Control and Health Monitoring, 29(8), e2973.
Kim, S., & Singh, R. (2021). Attention-based temporal graph neural networks for anomaly detection in water pipeline systems. IEEE Internet of Things Journal, 8(18), 14671–14682.
Chen, L., Wang, Z., & Hu, J. (2024). Edge-feature enhanced dynamic GNNs for infrastructure monitoring with environmental stressors. Journal of Big Data, 11(1), 17.
Li, M., Zhao, Y., & Sun, X. (2023). Graph embedding approaches for early-stage corrosion prediction in underground infrastructure. Computers & Industrial Engineering, 179, 108019.
Ahmed, F., & Tran, D. (2022). Multi-scale graph clustering for regional infrastructure risk assessment using sensor telemetry. IEEE Access, 10, 81245–81259.
Liu, B., Han, S., & Wang, L. (2023). Real-time fault detection using graph temporal convolutional networks in smart infrastructure. Sensors, 23(4), 1723.
Nguyen, H., Chen, K., & Yao, W. (2023). Adaptive temporal attention mechanisms for structural health monitoring systems. Future Generation Computer Systems, 144, 652–662.
Rao, P., & Das, S. (2023). Graph-based anomaly detection using streaming sensor data in civil infrastructure. Pattern Recognition Letters, 167, 21–28.
Song, J., Tan, Z., & Chatterjee, M. (2023). Scalable distributed training of graph neural networks for infrastructure monitoring. Cluster Computing, 26(3), 2459–2473.
Jain, A., Lee, J., & Zhao, Q. (2023). Building temporal graph data pipelines for predictive maintenance in smart infrastructure. Journal of Industrial Information Integration, 30, 100397.
Park, S., & Choi, M. (2023). Comparative study of time-series and graph-based models in fault prediction for sensor-driven systems. IEEE Transactions on Industrial Electronics, 70(7), 6843–6852.
Downloads
Published
Issue
Section
License
Copyright (c) 2025 Nicolas Thomas (Author)
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
