TY - GEN
T1 - Study on Inter-turn Short Circuit Fault Diagnosis Methods for AC Permanent Magnet Synchronous Motors
AU - Zhang, Lu
AU - Ma, Jie
AU - Guo, Wei
AU - Zhang, Yuanrui
N1 - Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026.
PY - 2026
Y1 - 2026
N2 - Permanent Magnet Synchronous Motors (PMSMs) are critical components in industrial applications, making accurate fault diagnosis essential for system health management. This paper focuses on the early diagnosis of inter-turn short circuit (ITSC) faults. A fault simulation model is established in the d-q coordinate system based on the mathematical principles of PMSMs, from which the q-axis current, voltage, torque, and speed signals are extracted as potential features. A hybrid RF-OOB-LDA-RF diagnostic framework is proposed: time-domain, frequency-domain, and time-frequency features are first extracted from simulation data under noisy conditions. A Random Forest (RF) algorithm, integrated with Out-of-Bag (OOB) estimation and Grid Search, is then employed for optimal feature selection. Linear Discriminant Analysis (LDA) further reduces the dimensionality of the selected feature set, which is finally fed into another RF classifier. This approach enhances both diagnostic accuracy and model generalization. Experimental results from a physical motor simulation platform validate the method’s effectiveness and feasibility for real-world systems, providing crucial technical support for PMSM health management.
AB - Permanent Magnet Synchronous Motors (PMSMs) are critical components in industrial applications, making accurate fault diagnosis essential for system health management. This paper focuses on the early diagnosis of inter-turn short circuit (ITSC) faults. A fault simulation model is established in the d-q coordinate system based on the mathematical principles of PMSMs, from which the q-axis current, voltage, torque, and speed signals are extracted as potential features. A hybrid RF-OOB-LDA-RF diagnostic framework is proposed: time-domain, frequency-domain, and time-frequency features are first extracted from simulation data under noisy conditions. A Random Forest (RF) algorithm, integrated with Out-of-Bag (OOB) estimation and Grid Search, is then employed for optimal feature selection. Linear Discriminant Analysis (LDA) further reduces the dimensionality of the selected feature set, which is finally fed into another RF classifier. This approach enhances both diagnostic accuracy and model generalization. Experimental results from a physical motor simulation platform validate the method’s effectiveness and feasibility for real-world systems, providing crucial technical support for PMSM health management.
KW - fault diagnosis
KW - Linear Discriminant Analysis
KW - permanent magnet synchronous motor
KW - random forest
KW - turn-to-turn short circuit
UR - https://www.scopus.com/pages/publications/105042988120
U2 - 10.1007/978-981-95-8435-2_56
DO - 10.1007/978-981-95-8435-2_56
M3 - 会议稿件
AN - SCOPUS:105042988120
SN - 9789819584345
T3 - Lecture Notes in Electrical Engineering
SP - 695
EP - 707
BT - Proceedings of 2025 9th Chinese Conference on Swarm Intelligence and Cooperative Control - Swarm Control Technologies
A2 - Wang, Qing
A2 - Dong, Xiwang
A2 - Song, Peng
PB - Springer Science and Business Media Deutschland GmbH
T2 - 9th Chinese Conference on Swarm Intelligence and Cooperative Control, CCSICC 2025
Y2 - 31 October 2025 through 3 November 2025
ER -