Data-Driven Predictive Maintenance of Permanent Magnet Synchronous Motor Systems Using Explainable Machine Learning and Digital Twin Architecture
DOI:
https://doi.org/10.65455/30tg2482Keywords:
Predictive Maintenance, Permanent Magnet Synchronous Motor, Explainable Machine Learning, SHAP, Digital Twin, Industry 4.0Abstract
Permanent magnet synchronous motors (PMSMs) are widely used in industrial automation, robotics, electric drives, and intelligent manufacturing, yet their maintenance still depends heavily on threshold alarms and periodic inspection. This paper presents a data-driven predictive maintenance framework for PMSM systems by integrating explainable machine learning with a digital twin architecture. A self-consistent synthetic PMSM digital twin dataset is constructed to represent thermal, electrical, mechanical, control-related, and aging degradation signals, including winding temperature, vibration RMS, current harmonic distortion, q-axis current ripple, DC-link voltage ripple, speed tracking error, flux-linkage index, insulation resistance, load ratio, and operating hours. Random Forest, XGBoost, and LightGBM are trained to classify the motor health state into healthy, warning, and fault conditions. The results show that LightGBM achieves the best held-out performance, with 95.00% accuracy, 0.9256 macro F1-score, and 0.9914 macro ROC-AUC. A 5-fold stratified cross-validation experiment confirms the stability of the model, while an ablation study verifies the contribution of thermal, mechanical, electrical, control, and aging feature groups. An additional AI4I 2020 benchmark validation is included to test whether the modeling pipeline generalizes to a public predictive-maintenance dataset. SHAP analysis indicates that winding temperature, vibration RMS, insulation resistance, speed error, and current harmonic distortion are dominant contributors to fault prediction. Finally, a deployable digital twin architecture is proposed for automation scenarios. The study provides a reproducible simulation-based reference for intelligent motor-drive maintenance under Industry 4.0.
References
[1]JARDINE A K S, LIN D, BANJEVIC D. A review on machinery diagnostics and prognostics implementing condition-based maintenance. Mechanical Systems and Signal Processing, 2006, 20(7): 1483-1510. DOI: https://doi.org/10.1016/j.ymssp.2005.09.012
[2]LEE J, BAGHERI B, KAO H A. A cyber-physical systems architecture for Industry 4.0-based manufacturing systems. Manufacturing Letters, 2015, 3: 18-23. DOI: https://doi.org/10.1016/j.mfglet.2014.12.001
[3]DALZOCHIO J, RODRIGUES R, CALDAS M, et al. Machine learning and reasoning for predictive maintenance in Industry 4.0: Current status and challenges. Computers in Industry, 2020, 123: 103298. DOI: https://doi.org/10.1016/j.compind.2020.103298
[4]BREIMAN L. Random forests. Machine Learning, 2001, 45(1): 5-32. DOI: https://doi.org/10.1023/A:1010933404324
[5]CHEN T, GUESTRIN C. XGBoost: A scalable tree boosting system. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2016: 785-794.
[6]KE G, MENG Q, FINLEY T, et al. LightGBM: A highly efficient gradient boosting decision tree. Advances in Neural Information Processing Systems, 2017, 30: 3146-3154.
[7]LUNDBERG S M, LEE S I. A unified approach to interpreting model predictions. Advances in Neural Information Processing Systems, 2017, 30: 4765-4774.
[8]GRIEVES M, VICKERS J. Digital twin: Mitigating unpredictable, undesirable emergent behavior in complex systems. In: Kahlen F J, Flumerfelt S, Alves A, eds. Transdisciplinary Perspectives on Complex Systems. Cham: Springer, 2017: 85-113. DOI: https://doi.org/10.1007/978-3-319-38756-7_4
[9]KRITZINGER W, KARNER M, TRAAR G, et al. Digital Twin in manufacturing: A categorical literature review and classification. IFAC-PapersOnLine, 2018, 51(11): 1016-1022. DOI: https://doi.org/10.1016/j.ifacol.2018.08.474
[10]MATZKA S. AI4I 2020 Predictive Maintenance Dataset. UCI Machine Learning Repository, 2020.
[11]SAXENA A, GOEBEL K. Turbofan Engine Degradation Simulation Data Set. NASA Ames Prognostics Data Repository, 2008.
[12]DERECI U, TUZKAYA G. An explainable artificial intelligence model for predictive maintenance and spare parts optimization. Supply Chain Analytics, 2024, 8: 100078. DOI: https://doi.org/10.1016/j.sca.2024.100078
[13]MALLIORIS P, AIVAZIDOU E, BECHTSIS D. Predictive maintenance in Industry 4.0: A systematic multi-sector mapping. CIRP Journal of Manufacturing Science and Technology, 2024, 50: 80-103. DOI: https://doi.org/10.1016/j.cirpj.2024.02.003
[14]MOBLEY R K. An Introduction to Predictive Maintenance. 2nd ed. New York: Butterworth-Heinemann, 2002. DOI: https://doi.org/10.1016/B978-075067531-4/50006-3
[15]ACHOUCH M, DIMITROVA M, ZAIDI S, et al. On predictive maintenance in Industry 4.0: Overview, models, and challenges. Applied Sciences, 2022, 12(16): 8081. DOI: https://doi.org/10.3390/app12168081
[16]KUMAR N B, BABU A R V, KUMAR M B A, KUMAR T S, BABU V G. Hybrid digital twin-based fault diagnosis framework for PMSMs in electric vehicle applications. Franklin Open, 2025, 12: 100328. DOI: https://doi.org/10.1016/j.fraope.2025.100328
[17]GHERGHINA I S, BIZON N, IANA G V, VASILICA B V. Recent advances in fault detection and analysis of synchronous motors: A review. Machines, 2025, 13(9): 815. DOI: https://doi.org/10.3390/machines13090815
[18]CHEN R, LIN S. Digital-twin-driven PMSM inter-turn short-circuit fault diagnosis method. Energies, 2026, 19(5): 1152. DOI: https://doi.org/10.3390/en19051152
[19]XIE T W, ZHANG X K, LIU X D, ZHANG X. Research on performance prediction model of wind turbine gearbox lubricating oil based on deep learning. Applied Artificial Intelligence Research, 2026, 2(1).
[20]KANG H, CHEN J, LIANG S, LIU S. An AI-driven intelligent health assessment system for high-voltage switchgear based on multimodal sensor data fusion. Applied Artificial Intelligence Research, 2026, 2(1). DOI: https://doi.org/10.65455/a6eqkh55
[21]LIU Z. Explainable machine learning for telecom customer churn prediction and actionable retention strategies. Applied Artificial Intelligence Research, 2026, 2(1). DOI: https://doi.org/10.65455/1z4v1627
[22]EL-DALAHMEH M, ABDULHADI N, COX S M, et al. Autonomous fault detection and diagnosis for permanent magnet synchronous motors using stator current signatures. Computers and Electrical Engineering, 2023, 109: 108785.
[23]PIETRZAK P, WOLKIEWICZ M. Demagnetization fault diagnosis of permanent magnet synchronous motors based on stator current signal processing and machine learning algorithms. Sensors, 2023, 23(4): 1757. DOI: https://doi.org/10.3390/s23041757
[24]XU X, HUANG W, CHEN Y, et al. Review of intelligent fault diagnosis for electric vehicle permanent magnet synchronous motors. Advances in Mechanical Engineering, 2020, 12(8): 1-17. DOI: https://doi.org/10.1177/1687814020944323
[25]GUO Z, PAN X. Research on fault diagnosis in the operation monitoring of permanent magnet synchronous motors through deep learning. Frontiers in Mechanical Engineering, 2025, 11: 1687802. DOI: https://doi.org/10.3389/fmech.2025.1687802
[26]RIBEIRO M T, SINGH S, GUESTRIN C. Why should I trust you? Explaining the predictions of any classifier. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2016: 1135-1144. DOI: https://doi.org/10.1145/2939672.2939778
[27]AIVALIOTIS P, GEORGOUDAKIS M, ARTOPOULOS A, et al. The use of Digital Twin for predictive maintenance in manufacturing. International Journal of Computer Integrated Manufacturing, 2019, 32(11): 1067-1080. DOI: https://doi.org/10.1080/0951192X.2019.1686173
[28]VAN DINTER R, TEKINERDOGAN B, CATAL C. Predictive maintenance using digital twins: A systematic literature review. Information and Software Technology, 2022, 151: 107008. DOI: https://doi.org/10.1016/j.infsof.2022.107008
Downloads
Published
Data Availability Statement
All data generated and analyzed for the PMSM study are synthetic and are described in the article. The dataset was produced by the simulation procedure reported in Section 3 and is intended for reproducible methodological evaluation rather than as a measured factory dataset. The AI4I 2020 dataset used for the additional benchmark validation is publicly available from the UCI Machine Learning Repository.
Issue
Section
License
Copyright (c) 2026 The Author(s). Applied Artificial Intelligence Research published by CSTDP
This work is licensed under a Creative Commons Attribution 4.0 International License.
