A Cyber Physical Framework for Intelligent Fault Detection and Predictive Maintenance in Digital Power Grids
Abstract
INTRODUCTION: The increasing complexity of modern digital power grids demands intelligent and reliable fault detection and predictive maintenance solutions to ensure operational stability, resilience, and efficient asset management. Integrating cyber–physical systems with advanced deep learning techniques enables accurate monitoring and proactive maintenance of power transmission networks. OBJECTIVES: This study proposes a unified GRU-based cyber–physical framework for intelligent fault detection, fault classification, fault localization, and risk-based predictive maintenance in power transmission systems, while improving diagnostic accuracy, interpretability, and maintenance decision-making. METHODS: The proposed framework integrates multi-parametric voltage and current measurements from the physical layer with a gated recurrent unit (GRU) model in the cyber layer to capture the temporal dynamics of fault evolution. Domain-informed electrical features, including voltage deviation, phase imbalance index, and current severity index, are incorporated to improve physical interpretability. A multi-task learning strategy is employed to simultaneously detect faults, classify fault types, identify faulted transmission lines, estimate fault locations, and generate a risk-based maintenance index for proactive maintenance prioritization. RESULTS: The proposed framework achieved a fault classification accuracy of 98.61%, with precision of 0.9864, recall of 0.9861, and F1-score of 0.9861. Fault localization was performed with a low mean absolute error (MAE) of 0.0303, demonstrating high localization accuracy. The risk-based maintenance index effectively supported proactive maintenance prioritization for transmission assets. CONCLUSION: The proposed GRU-based cyber–physical framework provides an effective and scalable solution for intelligent power grid monitoring by integrating accurate fault detection, precise fault localization, and predictive maintenance within a unified architecture. The framework enhances operational reliability and supports data-driven maintenance strategies for next-generation digital power grids.
Related
- Epanechnikov Pearson Type V distribution for reliability and predictive maintenance modeling
- Transformer-aided guard: A hybrid spatio-temporal framework for confidence-aware intrusion detection in cyber-physical systems
- Time-Invariant Learning and History-Based Inference for Survival Models in Predictive Maintenance
- Causal DAG-Guided Reinforcement Learning for Long-Term Physical Activity Maintenance: Using Prior Data to Inform Future Deployment
- A multi-stage framework for emotionally intelligent dialogue systems using topological sentiment encoding and causal policy optimization
BibTeX
@article{chen2026cyber,
title = {{A Cyber Physical Framework for Intelligent Fault Detection and Predictive Maintenance in Digital Power Grids}},
author = {Zhuangwei Chen and Beilei Zhang and Qing Jiang and Wenyong Zhong and Haohao Fu},
journal = {EAI Endorsed Transactions on Energy Web},
year = {2026},
month = oct,
doi = {10.4108/ew.13931},
url = {https://doi.org/10.4108/ew.13931},
note = {CC BY-NC-SA},
}
RIS
TY - JOUR
TI - A Cyber Physical Framework for Intelligent Fault Detection and Predictive Maintenance in Digital Power Grids
AU - Zhuangwei Chen
AU - Beilei Zhang
AU - Qing Jiang
AU - Wenyong Zhong
AU - Haohao Fu
PY - 2026
DA - 2026/10/23
JO - EAI Endorsed Transactions on Energy Web
DO - 10.4108/ew.13931
UR - https://doi.org/10.4108/ew.13931
AB - INTRODUCTION: The increasing complexity of modern digital power grids demands intelligent and reliable fault detection and predictive maintenance solutions to ensure operational stability, resilience, and efficient asset management. Integrating cyber–physical systems with advanced deep learning techniques enables accurate monitoring and proactive maintenance of power transmission networks. OBJECTIVES: This study proposes a unified GRU-based cyber–physical framework for intelligent fault detection, fault classification, fault localization, and risk-based predictive maintenance in power transmission systems, while improving diagnostic accuracy, interpretability, and maintenance decision-making. METHODS: The proposed framework integrates multi-parametric voltage and current measurements from the physical layer with a gated recurrent unit (GRU) model in the cyber layer to capture the temporal dynamics of fault evolution. Domain-informed electrical features, including voltage deviation, phase imbalance index, and current severity index, are incorporated to improve physical interpretability. A multi-task learning strategy is employed to simultaneously detect faults, classify fault types, identify faulted transmission lines, estimate fault locations, and generate a risk-based maintenance index for proactive maintenance prioritization. RESULTS: The proposed framework achieved a fault classification accuracy of 98.61%, with precision of 0.9864, recall of 0.9861, and F1-score of 0.9861. Fault localization was performed with a low mean absolute error (MAE) of 0.0303, demonstrating high localization accuracy. The risk-based maintenance index effectively supported proactive maintenance prioritization for transmission assets. CONCLUSION: The proposed GRU-based cyber–physical framework provides an effective and scalable solution for intelligent power grid monitoring by integrating accurate fault detection, precise fault localization, and predictive maintenance within a unified architecture. The framework enhances operational reliability and supports data-driven maintenance strategies for next-generation digital power grids.
SN - 2032-944X
ER -