Federated Learning-Based Edge-Cloud Collaborative Framework for Real-Time PMSM Fault Diagnosis
Keywords:
Federated learning, PMSM, Fault diagnosis, Edge computing, Cloud collaborationAbstract
Permanent magnet synchronous motors are critical components in modern industrial automation, electric transportation, and renewable energy systems, where operational reliability directly impacts safety, efficiency, and downtime costs. Traditional fault diagnosis approaches rely on centralized data collection and cloud-based analytics, which introduce latency, bandwidth bottlenecks, and privacy vulnerabilities-especially when deployed across geographically distributed edge devices. This paper presents a federated learning-based edge-cloud collaborative framework specifically engineered for real-time fault diagnosis of permanent magnet synchronous motors. The architecture decouples model training from raw data movement: local edge nodes perform on-device feature extraction and lightweight model updates using localized motor current, vibration, and thermal signatures, while a central cloud orchestrator aggregates and refines global model parameters without accessing raw sensor streams. A time-synchronized gradient averaging protocol ensures convergence stability under heterogeneous device capabilities, intermittent connectivity, and non-IID fault data distributions. Real-world validation was conducted across twelve industrial motor testbeds operating under variable load, speed, and ambient conditions. Results demonstrate sub-second diagnostic latency, 98.7% average classification accuracy across seven fault classes-including demagnetization, inter-turn short circuits, and bearing degradation-and a 73% reduction in uplink data volume compared to conventional cloud-only pipelines. The framework maintains robustness during network partitions and adapts continuously as new fault patterns emerge at the edge. By harmonizing computational responsibility between resource-constrained endpoints and scalable cloud infrastructure, this work advances a practical, privacy-aware, and responsive paradigm for intelligent motor health monitoring in dynamic industrial ecosystems.References
1. J. Wu, "A Methodological Framework for Training Tool-Aware AI Entertainers for Live Streaming," Artificial Intelligence and Digital Technology, vol. 3, no. 3, pp. 1-12, 2026.
2. M. Jiang, H. Liang, C. Qian, and W. Yu, “Machine learning algorithms for forecasting and diagnosis applications in smart grids,” in Digital Twin Technology and Smart Grid. Amsterdam: Elsevier, 2026, pp. 103–124.
3. C. Zhang, J. Huang, H. Yu, W. Yin, X. Liang, and X. Luo, “As cities sprawl: Unraveling the nonlinear impacts of urban spatial evolution on outdoor crime patterns through machine learning and causal inference,” Sustain. Cities Soc., Art. no. 107475, 2026.
4. W. Zhang, M. Li, and X. Chen, “Distributed edge-fog-cloud architecture for intelligent fault diagnosis of permanent magnet synchronous motors using temporal fusion transformers and federated learning,” Fundam. Sci. Rep. Multidiscip. Areas, vol. 2, no. 02, pp. 121–131, 2026.
5. N. Tulupov and D. Kanurin, “Physics-informed TD3 for end-to-end torque control of a PMSM drive,” in Proc. 2026 IEEE 27th Int. Conf. Young Profess. Electron Devices Mater. (EDM), Jun. 2026, pp. 1–6.
6. J. Zhang and H. Gao, “LLM inference delay prediction based on machine learning algorithms,” in *Proc. 2026 Int. Conf. Generative Artificial Intelligence and Information Security (GAIIS)*, Mar. 2026, pp. 37–41.
7. S. Sun, B. Chen, H. Niu, Z. Zhang, X. Zhang, and Z. Song, “Rehearsal-free federated continual learning with prototype guided strategy for industrial fault diagnosis,” IEEE Sens. J., 2026.
8. W. Zhang, T. Liao, and Y. Feng, “Multi-Agent LLM Collaborative Reasoning and Task Planning for Complex Task Solving,” 2026. doi: 10.21203/rs.3.rs-10232223/v1.
9. F. Zheng and X. Feng, “A lightweight federated multitask fault diagnosis framework for PMSMs with knowledge distillation analysis,” in Proc. 2026 2nd Int. Conf. Smart Energy Smart Grid (SESG), Jun. 2026, pp. 254–258.
10. X. Fang, Y. H. Yang, and S. Wang, “Energy-Efficient Context-Aware Multimodal AI Inference at the Edge,” 2026.
11. K. Zhao, R. Zhou, J. She, C. Zhang, J. He, G. Huang, and X. Li, “Demagnetization-fault reconstruction and tolerant-control for PMSM using improved SMO-based equivalent-input-disturbance approach,” IEEE/ASME Trans. Mechatron., vol. 27, no. 2, pp. 701–712, 2021.
12. Y. Yan, J. Yang, Z. Sun, C. Zhang, S. Li, and H. Yu, “Robust speed regulation for PMSM servo system with multiple sources of disturbances via an augmented disturbance observer,” IEEE/ASME Trans. Mechatron., vol. 23, no. 2, pp. 769–780, 2018.
13. M. Ntombela and M. Kabeya, “A comprehensive review of artificial intelligence-driven health management of electrical machines,” 2026.
14. L. Wan, T. Sun, H. Tan, J. Ning, W. Ni, and K. Li, “A novel sensor heterogeneous federated domain generalization method based on prototype contrastive learning for intelligent fault diagnosis,” IEEE Internet Things J., 2026.
15. A. Mohammad-Alikhani, B. Nahid-Mobarakeh, and M. F. Hsieh, “One-dimensional LSTM-regulated deep residual network for data-driven fault detection in electric machines,” IEEE Trans. Ind. Electron., vol. 71, no. 3, pp. 3083–3092, 2023.
