Explainable AI for Modeling Magnetic Flux Distribution in MCCB Actuators: A Configuration-Aware Analysis
Keywords:
Explainable Artificial Intelligence (XAI), Magnetic Flux Prediction, MCCB, Magnetic Actuator, Feature Importance, Spatial Features, Surrogate Modeling, Random ForestAbstract
Background: This study presents an Explainable Artificial Intelligence (XAI)–based analysis of the three-dimensional magnetic flux density components (Bx, By, Bz) generated in magnetic actuators used in molded case circuit breakers (MCCBs). The analyses were conducted on a dataset obtained from electromagnetic simulations performed in CST Studio Suite, comprising a total of 1,186,066 samples across 27 different actuator configurations. The dataset includes 16 input features (Full Mode), covering design parameters, spatial coordinates, and physical field quantities.
Methods: To investigate how the model benefits from different levels of information, three feature modes were compared: (i) Full Mode, which includes all variables; (ii) Design–Spatial Mode, which incorporates design parameters and spatial coordinates; and (iii) Design-Only Mode, which relies solely on design parameters. To prevent data leakage, the evaluation was carried out using a configu-ration-based group splitting strategy, and model stability was assessed through repeated experiments.
Results: The results show that the Full Mode, as the most physically informative representation, achieves the highest accuracy, particularly for the By and Bz components. The Design–Spatial Mode, despite its simpler feature structure, largely preserves this high performance. In contrast, the Design-Only Mode exhibits a significant drop in performance. These findings highlight that spatial infor-mation is critical for the point-wise prediction of local magnetic flux density components, whereas relying solely on design parameters is insufficient. Furthermore, feature importance analysis and SHapley Additive exPlanations (SHAP)-based explanations indicate that the model’s decisions are grounded in physically meaningful variables. Quantitatively, the Full Mode attains R² = 0.9152, 0.9910, and 0.9941 for Bx, By, and Bz, respectively, while the more compact Design–Spatial Mode remains highly competitive (R² = 0.9846, 0.9916, and 0.9658), and the Design-Only Mode drops sharply (R² = 0.0194, 0.1015, and 0.0063 for Bx, By, and Bz respectively), confirming that spatial information — rather than design parameters alone — is the key driver of point-wise prediction accuracy.
Conclusion: Overall, the findings demonstrate that, in data-driven modeling of electromagnetic systems, not only predictive accuracy but also the level of information and interpretability play a crucial role.
References
Ali, S., Abuhmed, T., El-Sappagh, S., Muhammad, K., Alonso-Moral, J. M., Confalonieri, R., Guidotti, R., Del Ser, J., Díaz-Rodríguez, N. and Herrera, F. (2023), "Explainable Artificial Intelligence (XAI): What we know and what is left to attain Trustworthy Artificial Intelligence", Information Fusion, Vol. 99, pp. 101805.
Barredo Arrieta, A., Díaz-Rodríguez, N., Del Ser, J., Bennetot, A., Tabik, S., Barbado, A., Garcia, S., Gil-Lopez, S., Molina, D., Benjamins, R., Chatila, R. and Herrera, F. (2020), "Explainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI", Information Fusion, Vol. 58, pp. 82-115.
Breiman, L. (2001), "Random Forests", Machine Learning, Vol. 45 No. 1, pp. 5-32.
Chen, H., Li, X., Zhou, C., Tan, Y. and Shen, Y. (2025), "Simulation Study on Electromagnetic Response and Cable Coupling Characteristics of eVTOL Under Lightning Environment", Electronics, pp. 3661.
Cho, Y.-M., Park, H.-J., Lee, J.-J. and Lee, K.-A. (2022), "Analysis of Characteristics of Low Voltage Circuit Breaker by External Magnetic Field", Energies, pp. 8156.
Deng, Y., Fan, K., Jin, B., Malof, J. and Padilla, W. J. (2025), "Physics-informed learning in artificial electromagnetic materials", Applied Physics Reviews, Vol. 12 No. 1, pp. 011331.
Givisis, I., Kalatzis, D., Christakis, C. and Kiouvrekis, Y. (2025), "Comparing Explainable AI Models: SHAP, LIME, and Their Role in Electric Field Strength Prediction over Urban Areas", Electronics, pp. 4766.
Karniadakis, G. E., Kevrekidis, I. G., Lu, L., Perdikaris, P., Wang, S. and Yang, L. (2021), "Physics-informed machine learning", Nature Reviews Physics, Vol. 3 No. 6, pp. 422-440.
Kiouvrekis, Y., Givisis, I., Panagiotakopoulos, T., Tsilikas, I., Ploussi, A., Spyratou, E. and Efstathopoulos, E. P. (2025), "A Comparative Analysis of Explainable Artificial Intelligence Models for Electric Field Strength Prediction over Eight European Cities", Sensors, pp. 53.
Liu, L., Li, Z., Kang, H., Xiao, Y., Sun, L., Zhao, H., Zhu, Z. Q. and Ma, Y. (2025), "Review of surrogate model assisted multi-objective design optimization of electrical machines: New opportunities and challenges", Renewable and Sustainable Energy Reviews, Vol. 215, pp. 115609.
Liu, S., Jin, J., Chen, J., Yu, M., Liu, Z. and Geng, Y. (2024), "Evolution and interruption characteristics of DC vacuum arc under Halbach transverse magnetic fields", International Journal of Electrical Power & Energy Systems, Vol. 157, pp. 109844.
Lundberg, S. M., Erion, G., Chen, H., DeGrave, A., Prutkin, J. M., Nair, B., Katz, R., Himmelfarb, J., Bansal, N. and Lee, S.-I. (2020), "From local explanations to global understanding with explainable AI for trees", Nature Machine Intelligence, Vol. 2 No. 1, pp. 56-67.
Lundberg, S. M. and Lee, S.-I. (2017), "A unified approach to interpreting model predictions", in Proceedings of the 31st International Conference on Neural Information Processing Systems. Curran Associates Inc., Long Beach, California, USA, pp. 4768–4777.
Peng, R., Ren, S., Malof, J. and Padilla, W. J. (2024), "Transfer learning for metamaterial design and simulation", Vol. 13 No. 13, pp. 2323-2334.
Qoria, T., Gruson, F., Colas, F., Kestelyn, X. and Guillaud, X. (2020), "Current limiting algorithms and transient stability analysis of grid-forming VSCs", Electric Power Systems Research, Vol. 189, pp. 106726.
Rudin, C. (2019), "Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead", Nature Machine Intelligence, Vol. 1 No. 5, pp. 206-215.
Tjoa, E. and Guan, C. (2021), "A Survey on Explainable Artificial Intelligence (XAI): Toward Medical XAI", IEEE Transactions on Neural Networks and Learning Systems, Vol. 32 No. 11, pp. 4793-4813.
Wang, J., Wang, D., Wang, S., Li, W. and Jiang, Y. (2025), "Dimensionless Physics-Informed Neural Network for Electromagnetic Field Modelling of Permanent Magnet Eddy Current Coupler", IET Electric Power Applications, Vol. 19 No. 1, pp. e70084.
Peng, R., Ren, S., Malof, J., & Padilla, W. J. (2024). Transfer learning for metamaterial design and simulation. Nanophotonics, 13(13), 2323–2334.
Wang, Q., Ge, S., Cao, W., Yu, S. and Liao, Z. (2023), "Study of Arc Interruption Characteristics under Rated Current in Low Voltage Circuit Breakers", Energies, pp. 4114.
Wu, Y., Sicard, B. and Gadsden, S. A. (2024), "Physics-informed machine learning: A comprehensive review on applications in anomaly detection and condition monitoring", Expert Systems with Applications, Vol. 255, pp. 124678.
Zha, D., Bhat, Z. P., Lai, K.-H., Yang, F., Jiang, Z., Zhong, S. and Hu, X. (2025), "Data-centric Artificial Intelligence: A Survey", Vol. 57 No. 5 ACM Computing Surveys., pp. Article 129.
Zhang, B., Zhang, J., Cheng, Y., Chen, Q. and Zhang, Q. (2025), "Simulation and Experimental Study of Arc Model in a Low-Voltage Distribution Network", Energies, pp. 420.
Downloads
Published
Issue
Section
Categories
License
Copyright (c) 2026 Mehmet Onur YAĞIR, Sevda GÜL (Author)
This work is licensed under a Creative Commons Attribution 4.0 International License.
