XFed-PQD: Explainable Federated Learning for Power Quality Disturbance Classification in Edge-Enabled Smart Distribution Networks
DOI:
https://doi.org/10.65405/rd7zdr58الكلمات المفتاحية:
power quality disturbances; federated learning; FedAvg; FedProx; explainable artificial intelligence; Integrated Gradients; Grad-CAM; edge intelligence; dynamic quantization; non-IID dataالملخص
Power-quality disturbance (PQD) recognition is increasingly being pushed toward the sensing point rather than handled only at a central server. That shift creates a useful opportunity for distributed learning, but it also exposes two practical questions: how much performance is lost when raw waveforms remain local, and whether the resulting decisions can still be interpreted. This paper examines both questions through XFed-PQD, an explainable federated-learning framework for 17 single and compound PQD categories in edge-enabled smart distribution networks. The framework combines a compact multi-branch one-dimensional convolutional network with FedAvg and FedProx training, temporal attribution, robustness testing, and deployment profiling. The controlled experiment used three seeds, 17,000 reference waveforms, ten fully participating clients, 15 centralized epochs, and 25 federated rounds. Centralized training produced 91.35% ± 0.70% accuracy and 91.33% ± 0.70% macro-F1. With FedAvg and an IID partition, the same model reached 83.89% ± 0.78% accuracy and 83.85% ± 0.89% macro-F1 while exchanging 47.57 MB of model traffic, 83.4% less than the CNN-LSTM benchmark. Client heterogeneity had a much larger effect: Dirichlet label skew reduced macro-F1 to 69.41% ± 1.70%, and FedProx reached 69.67% ± 2.05%. The resulting 0.26-percentage-point average gain was small and did not point in the same direction for every seed. In the extended Seed-42 analysis, local-to-global Integrated-Gradients cosine similarity was 0.924 ± 0.053. The same checkpoint was highly sensitive to added noise, with macro-F1 falling from 83.82% on clean data to 54.97%, 17.01%, and 3.71% at 40, 30, and 20 dB. Dynamic INT8 quantization changed macro-F1 by -0.13 percentage points, reduced serialized size by 3.44%, and increased mean CPU latency. The dataset audit also uncovered a repeated healthy waveform that crossed data splits and appeared once under the Swell label; for that reason, three-seed sensitivity results excluding the true healthy class are reported. Taken together, the results show that explainable federated PQD monitoring is feasible under the evaluated conditions, while also making clear that client heterogeneity, duplicated waveforms, noise sensitivity, and hardware-dependent runtime behavior remain central concerns.
التنزيلات
المراجع
[1] B. McMahan, E. Moore, D. Ramage, S. Hampson, and B. A. y Arcas, “Communication-efficient learning of deep networks from decentralized data,” in Proc. 20th Int. Conf. Artificial Intelligence and Statistics (AISTATS), 2017, pp. 1273–1282.
[2] T. Li, A. K. Sahu, M. Zaheer, M. Sanjabi, A. Talwalkar, and V. Smith, “Federated optimization in heterogeneous networks,” in Proc. Machine Learning and Systems (MLSys), vol. 2, 2020, pp. 429–450.
[3] M. N. Fekri, K. Grolinger, and S. Mir, “Distributed load forecasting using smart meter data: Federated learning with recurrent neural networks,” Int. J. Electr. Power Energy Syst., vol. 137, Art. no. 107669, 2022, doi: 10.1016/j.ijepes.2021.107669.
[4] X. Cheng, C. Li, and X. Liu, “A review of federated learning in energy systems,” in Proc. IEEE/IAS Industrial and Commercial Power System Asia (I&CPS Asia), 2022, pp. 2089–2095, doi: 10.1109/ICPSAsia55496.2022.9949863.
[5] M. N. Fekri, K. Grolinger, and S. Mir, “Asynchronous adaptive federated learning for distributed load forecasting with smart meter data,” Int. J. Electr. Power Energy Syst., vol. 153, Art. no. 109285, 2023, doi: 10.1016/j.ijepes.2023.109285.
[6] N. Gholizadeh and P. Musilek, “Distributed learning applications in power systems: A review of methods, gaps, and challenges,” Energies, vol. 14, no. 12, Art. no. 3654, 2021, doi: 10.3390/en14123654.
[7] H. Bousbiat, R. Bousselidj, Y. Himeur, A. Amira, F. Bensaali, F. Fadli, W. Mansoor, and W. Elmenreich, “Crossing roads of federated learning and smart grids: Overview, challenges, and perspectives,” arXiv:2304.08602, 2023, doi: 10.48550/arXiv.2304.08602.
[8] R. Zheng, A. Sumper, M. Aragüés-Peñalba, and S. Galceran-Arellano, “Advancing power system services with privacy-preserving federated learning techniques: A review,” IEEE Access, vol. 12, pp. 76753–76780, 2024, doi: 10.1109/ACCESS.2024.3407121.
[9] Z. Zhang, S. Rath, J. Xu, and T. Xiao, “Federated learning for smart grid: A survey on applications and potential vulnerabilities,” ACM Comput. Surv., 2026, doi: 10.1145/3760788.
[10] R. Zhao, J. Lu, Z. Liu, T. Wang, W. Guo, T. Lan, and C. Hu, “A federated-learning algorithm based on client sampling and gradient projection for the smart grid,” Electronics, vol. 13, no. 11, Art. no. 2023, 2024, doi: 10.3390/electronics13112023.
[11] H. Lee, “Towards convergence in federated learning via non-IID analysis in a distributed solar energy grid,” Electronics, vol. 12, no. 7, Art. no. 1580, 2023, doi: 10.3390/electronics12071580.
[12] R. Rahman, N. Kumar, and D. C. Nguyen, “Electrical load forecasting in smart grid: A personalized federated learning approach,” arXiv:2411.10619, 2024, doi: 10.48550/arXiv.2411.10619.
[13] S. Ghosh and G. Mittal, “Federated learning for critical electrical infrastructure—handling data heterogeneity for predictive maintenance of substation equipment,” Front. Artif. Intell., vol. 8, Art. no. 1697175, 2026, doi: 10.3389/frai.2025.1697175.
[14] A. Tibermacine, I. Naidji, I. E. Tibermacine, S. Sahraoui, M. Zouai, A. Rabehi, and M. Habib, “Privacy-preserving load forecasting in smart grids using federated learning: A comparative analysis of aggregation strategies,” Front. Energy Res., vol. 14, Art. no. 1854654, 2026, doi: 10.3389/fenrg.2026.1854654.
[15] M. A. Husnoo, A. Anwar, N. Hosseinzadeh, S. N. Islam, A. N. Mahmood, and R. Doss, “A secure federated learning framework for residential short-term load forecasting,” IEEE Trans. Smart Grid, vol. 15, no. 2, pp. 2044–2055, 2024, doi: 10.1109/TSG.2023.3292382.
[16] M. A. Husnoo, A. Anwar, H. T. Reda, N. Hosseinzadeh, S. N. Islam, A. N. Mahmood, and R. Doss, “FedDiSC: A computation-efficient federated learning framework for power systems disturbance and cyber attack discrimination,” Energy and AI, vol. 14, Art. no. 100271, 2023, doi: 10.1016/j.egyai.2023.100271.
[17] J. Jithish, B. Alangot, N. Mahalingam, and K. S. Yeo, “Distributed anomaly detection in smart grids: A federated learning-based approach,” IEEE Access, vol. 11, pp. 7157–7179, 2023, doi: 10.1109/ACCESS.2023.3237554.
[18] M. Kesici, B. Pal, and G. Yang, “Detection of false data injection attacks in distribution networks: A vertical federated learning approach,” IEEE Trans. Smart Grid, vol. 15, no. 6, pp. 5952–5964, 2024, doi: 10.1109/TSG.2024.3399396.
[19] Y. Li, D. Qin, H. V. Poor, and Y. Wang, “Introducing edge intelligence to smart meters via federated split learning,” Nat. Commun., vol. 15, Art. no. 9044, 2024, doi: 10.1038/s41467-024-53352-9.
[20] N. Hudson, M. J. Hossain, M. Hosseinzadeh, H. Khamfroush, M. Rahnamay-Naeini, and N. Ghani, “A framework for edge intelligent smart distribution grids via federated learning,” in Proc. 30th Int. Conf. Computer Communications and Networks (ICCCN), 2021, doi: 10.1109/ICCCN52240.2021.9522360.
[21] D. N. Molokomme, A. J. Onumanyi, and A. M. Abu-Mahfouz, “Edge intelligence in smart grids: A survey on architectures, offloading models, cyber security measures, and challenges,” J. Sens. Actuator Netw., vol. 11, no. 3, Art. no. 47, 2022, doi: 10.3390/jsan11030047.
[22] D. Liu, H. Liang, X. Zeng, Q. Zhang, Z. Zhang, and M. Li, “Edge computing application, architecture, and challenges in ubiquitous power Internet of Things,” Front. Energy Res., vol. 10, Art. no. 850252, 2022, doi: 10.3389/fenrg.2022.850252.
[23] W. Li, N. Zhang, Z. Liu, S. Ma, H. Ke, J. Wang, and T. Chen, “A trusted decision fusion approach for the power Internet of Things with federated learning,” Front. Energy Res., vol. 11, Art. no. 1061779, 2023, doi: 10.3389/fenrg.2023.1061779.
[24] H. Gupta, P. Agarwal, K. Gupta, S. Baliarsingh, O. P. Vyas, and A. Puliafito, “FedGrid: A secure framework with federated learning for energy optimization in the smart grid,” Energies, vol. 16, no. 24, Art. no. 8097, 2023, doi: 10.3390/en16248097.
[25] B. Saylam and Ö. D. İncel, “Federated learning on edge sensing devices: A review,” arXiv:2311.01201, 2023, doi: 10.48550/arXiv.2311.01201.
[26] R. Machlev, A. Chachkes, J. Belikov, Y. Beck, and Y. Levron, “Open source dataset generator for power quality disturbances with deep-learning reference classifiers,” Electr. Power Syst. Res., vol. 195, Art. no. 107152, 2021, doi: 10.1016/j.epsr.2021.107152.
[27] M. Alsabaan, A. Elsayed, A. Bondok, M. M. Badr, M. Mahmoud, T. Alshawi, and M. I. Ibrahem, “Robust federated-learning-based classifier for smart grid power quality disturbances,” Sensors, vol. 25, no. 22, Art. no. 6880, 2025, doi: 10.3390/s25226880.
[28] A. M. Saber, A. Selim, M. M. Hammad, A. Youssef, D. Kundur, and E. El-Saadany, “A novel approach to classify power quality signals using vision transformers,” in Proc. 50th Annual Conf. IEEE Industrial Electronics Society (IECON), 2024, doi: 10.1109/IECON55916.2024.10905293.
[29] S. Yang, T. Shan, and X. Yan, “Interpretable DWT-1DCNN-LSTM network for power quality disturbance classification,” Energies, vol. 18, no. 2, Art. no. 231, 2025, doi: 10.3390/en18020231.
[30] H. Bai, R. Yao, T. Liu, Z. Ma, S. Liu, Y. Lei, and Y. Zheng, “A lightweight model for power quality disturbance recognition targeting edge deployment,” Energies, vol. 19, no. 2, Art. no. 368, 2026, doi: 10.3390/en19020368.
[31] H. Bai, R. Yao, W. Zhang, Z. Zhong, and H. Zou, “Power quality disturbance classification strategy based on fast S-transform and an improved CNN-LSTM hybrid model,” Processes, vol. 13, no. 3, Art. no. 743, 2025, doi: 10.3390/pr13030743.
[32] B. E. Altun, F. Alpsalaz, H. Uzel, and Y. Türkay, “Explainable DL based classification for power quality disturbances in renewable-energy-integrated distribution networks,” IET Renew. Power Gener., vol. 20, no. 1, Art. no. e70269, 2026, doi: 10.1049/rpg2.70269.
[33] Y. Chen, S. S. Yu, and K. M. Muttaqi, “A Bayesian framework for uncertainty-aware explanations in power quality disturbance classification,” arXiv:2604.13658, 2026, doi: 10.48550/arXiv.2604.13658.
[34] F. A. Albalooshi and M. R. Qader, “Deep learning algorithm for automatic classification of power quality disturbances,” Appl. Sci., vol. 15, no. 3, Art. no. 1442, 2025, doi: 10.3390/app15031442.
[35] C. A. Iturrino Garcia, M. Bindi, F. Corti, A. Luchetta, F. Grasso, L. Paolucci, M. C. Piccirilli, and I. Aizenberg, “Power quality analysis based on machine learning methods for low-voltage electrical distribution lines,” Energies, vol. 16, no. 9, Art. no. 3627, 2023, doi: 10.3390/en16093627.
[36] J. Cai, K. Zhang, and H. Jiang, “Power quality disturbance classification based on parallel fusion of CNN and GRU,” Energies, vol. 16, no. 10, Art. no. 4029, 2023, doi: 10.3390/en16104029.
[37] S. Samanta et al., “A comprehensive review of deep-learning applications to power quality analysis,” Energies, vol. 16, no. 11, Art. no. 4406, 2023, doi: 10.3390/en16114406.
[38] M. A. A. Baig, N. I. Ratyal, A. Amin, U. Jamil, S. Liaquat, H. M. Khalid, and M. F. Zia, “An ensemble deep CNN approach for power quality disturbance classification: A technological route towards smart cities using image-based transfer,” Future Internet, vol. 16, no. 12, Art. no. 436, 2024, doi: 10.3390/fi16120436.
[39] M. N. Islam, “A multimodal deep learning model with differential evolution-based optimized features for classification of power quality disturbances,” J. Electr. Syst. Inf. Technol., vol. 12, Art. no. 11, 2025, doi: 10.1186/s43067-025-00194-0.
[40] W.-M. Lin and C.-H. Wu, “Fast support vector machine for power quality disturbance classification,” Appl. Sci., vol. 12, no. 22, Art. no. 11649, 2022, doi: 10.3390/app122211649.
[41] S. Chamchuen, A. Siritaratiwat, P. Fuangfoo, P. Suthisopapan, and P. Khunkitti, “High-accuracy power quality disturbance classification using the adaptive ABC-PSO as optimal feature selection algorithm,” Energies, vol. 14, no. 5, Art. no. 1238, 2021, doi: 10.3390/en14051238.
[42] L. Gao, J. Wang, M. Zhang, S. Zhang, H. Wang, and Y. Wang, “Classification strategy for power quality disturbances based on variational mode decomposition algorithm and improved support vector machine,” Processes, vol. 12, no. 6, Art. no. 1084, 2024, doi: 10.3390/pr12061084.
[43] Q. Xu, F. Zhu, W. Jiang, X. Pan, P. Li, X. Zhou, and Y. Wang, “Efficient identification method for power quality disturbance: A hybrid data-driven strategy,” Processes, vol. 12, no. 7, Art. no. 1395, 2024, doi: 10.3390/pr12071395.
[44] B. F. Khaldi, F. Z. Dekhandji, and A. Recioui, “Low-cost IoT-based smart meter for real-time power quality monitoring and disturbance detection using embedded 1D CNN,” J. Energy Syst., vol. 9, no. 4, pp. 365–379, 2025, doi: 10.30521/jes.1718242.
[45] M. Ingram, R. Mahmud, and D. Narang, “Background information on the power quality requirements in IEEE Std 1547-2018,” National Renewable Energy Laboratory, Golden, CO, USA, Tech. Rep. NREL/TP-5D00-78751, 2021.
[46] M. Sundararajan, A. Taly, and Q. Yan, “Axiomatic attribution for deep networks,” in Proc. 34th Int. Conf. Machine Learning (ICML), 2017, pp. 3319–3328.
[47] R. R. Selvaraju et al., “Grad-CAM: Visual explanations from deep networks via gradient-based localization,” in Proc. IEEE Int. Conf. Computer Vision (ICCV), 2017, pp. 618–626, doi: 10.1109/ICCV.2017.74.
[48] A. Hedström et al., “Quantus: An explainable AI toolkit for responsible evaluation of neural network explanations and beyond,” J. Mach. Learn. Res., vol. 24, no. 34, pp. 1–11, 2023.
[49] Alriheebi, R. A. (2026). Integrated Modeling and Mult objective Control of Multilevel DC/AC Inverters for Conducted Electromagnetic Interference Mitigation. Al-Farooq Journal of Sciences, 2(3), 897-925.
[50] M. Nagel, M. Fournarakis, R. A. Amjad, Y. Bondarenko, M. van Baalen, and T. Blankevoort, “A white paper on neural network quantization,” arXiv:2106.08295, 2021, doi: 10.48550/arXiv.2106.08295.
[51] R. Alsaigh, R. Mehmood, and I. Katib, “AI explainability and governance in smart energy systems: A review,” Front. Energy Res., vol. 11, Art. no. 1071291, 2023, doi: 10.3389/fenrg.2023.1071291.
