基于领域对抗网络的永磁同步电机初始位置估计

doi:10.19595/j.cnki.1000-6753.tces.232019

Abstract
Figure/Table
References
Related Citation (8)

Download: PDF (6651 KB) HTML (1 KB)
Export: BibTeX | EndNote (RIS)

Abstract The low-speed sensorless control of the permanent magnet synchronous motor (PMSM) primarily depends on the motor's saliency characteristics. However, the surface-mounted permanent magnet synchronous motor (SPMSM) exhibits a low inductive saliency ratio, leading to a poor signal-to-noise ratio (SNR) in position signals. It challenges traditional high-frequency injection position estimation models in achieving high-precision estimation. Deep machine learning can extract features from subtle signals and accurately determine position even in low SNR environments. However, traditional supervised machine learning methods require abundant labeled training data. In practical applications, instantaneous load variations make it challenging to collect comprehensive samples and accurately label position tags, thus impacting the generalization ability of position estimation models trained on steady-state datasets. This paper proposes a data domain-adversarial neural network (DANN). This method enhances the traditional convolutional neural network (CNN) by introducing a domain classifier and a gradient reversal layer (GRL), transitioning from a single-stream to a dual-stream structure. The two networks learn invariant domain features, adapt data distribution, and mitigate the degraded position estimation by anti-transfer learning.
Balanced three-phase high-frequency excitation voltages are applied to the three-phase windings of the motor in the stationary reference frame. The negative-sequence first-order component and the positive-sequence second-harmonic component of the high-frequency response current are combined to form a new current vector, facilitating the extraction of position and polarity information. These reconstructed current vectors are graphically depicted and employed as inputs for the CNN. Implicit position feature information is extracted from two-dimensional images, and a correspondence between the images and the rotor positions is established through training. Consequently, the CNN model accurately recognizes two-dimensional images for position estimation. This CNN network is used as the feature extractor in the DANN network, incorporating a domain discriminator to distinguish input features from the source and target domains. Furthermore, a gradient reversal layer (GRL) is introduced between the feature extractor and the domain classifier. In backpropagation, the gradient parameters of the loss function in the domain classifier are reversed, which facilitates adversarial learning and aligns the probability distribution of new test data with the training set.
Experimental results indicate that traditional observers have significant errors in position estimation, with numerous deviations exceeding 90°. The CNN-based position estimation model maintains an average positional deviation of 13.2°, whereas the DANN-based model further decreases this deviation to 8.6°. When predicting new, unfamiliar data, the DANN model exhibits notably superior position estimation performance compared to the CNN model. The dual-stream DANN model can precisely estimate the initial position of motors with low saliency ratios, and position estimation results on new data are improved. Consequently, the poor data generalization ability inherent in the single-stream CNN position estimation model is mitigated.
The following conclusions are drawn. By leveraging the correlation between images and positions, image recognition-based position estimation methods enable accurate position determination through high-frequency current image recognition, solving the issue of the initial position estimation with low saliency ratios. Introducing the gradient reversal layer for domain adversarial learning facilitates data distribution-adaptive transfer learning. The position can be accurately estimated even with notable disparities in data probability distributions between test data and the training set. The dual-stream DANN model demonstrates superior domain generalization capability and enhanced robustness compared to the single-stream CNN network model. This approach improves the predictive performance of deep learning in scenarios of limited training samples and high labeling costs. A robust theoretical and practical framework for industrial machine-learning applications has been established.

Key words： Image recognition transfer learning network adversarial data-adaptive data generalization

Received: 04 December 2023

PACS:

TM614

	Service

	E-mail this article
	Add to my bookshelf
	Add to citation manager
	E-mail Alert
	RSS
	Articles by authors
	Wang Henghong
	Wang Jiyao
	Xu Wei
	Jin Zhen
	Hu Youkang

Cite this article:

Wang Henghong,Wang Jiyao,Xu Wei等. Initial Position Estimation of Surface Permanent Magnet Synchronous Motor Base on Domain-Adversarial Neural Networks[J]. Transactions of China Electrotechnical Society, 2025, 40(2): 425-438.

URL:

https://dgjsxb.ces-transaction.com/EN/10.19595/j.cnki.1000-6753.tces.232019 OR https://dgjsxb.ces-transaction.com/EN/Y2025/V40/I2/425

[1] Mai Zhiqin, Xiao Fei, Fu Kangzhuang, et al.HF pulsating carrier voltage injection method based on improved position error signal extraction strategy for PMSM position sensorless control[J]. IEEE Transa- ctions on Power Electronics, 2021, 36(8): 9348-9360.
[2] Zhang Xing, Li Haoyuan, Yang Shuying, et al.Improved initial rotor position estimation for PMSM drives based on HF pulsating voltage signal injection[J]. IEEE Transactions on Industrial Elec- tronics, 2018, 65(6): 4702-4713.
[3] 杨洛鸿, 刘侃, 胡伟, 等. 基于交轴磁链辨识的永磁同步电机位置传感器零位校正方法[J]. 电气工程学报, 2022, 17(4): 163-173.
Yang Luohong, Liu Kan, Hu Wei, et al.A q-axis flux linkage identification based zero-correction for position sensor of permanent magnet synchronous motor[J]. Journal of Electrical Engineering, 2022, 17(4): 163-173.
[4] Wu Lijian, Lü Zekai, Chen Zekai, et al.An enhanced sensorless control scheme for PMSM drives con- sidering self-inductance asymmetry[J]. CES Transa- ctions on Electrical Machines and Systems, 2022, 6(4): 384-392.
[5] Sun Qingguo, Zhu Xiaolei, Niu Feng.Sensorless control of permanent magnet synchronous motor based on new sliding mode observer with single resistor current reconstruction[J]. CES Transactions on Electrical Machines and Systems, 2022, 6(4): 378-383.
[6] 梅三冠, 卢闻州, 樊启高, 等. 基于滑模观测器误差补偿的永磁同步电机无位置传感器控制策略[J]. 电工技术学报, 2023, 38(2): 398-408.
Mei Sanguan, Lu Wenzhou, Fan Qigao, et al.Sensorless control strategy of permanent magnet synchronous motor based on error compensation estimated by sliding mode observer[J]. Transactions of China Electrotechnical Society, 2023, 38(2): 398-408.
[7] Xu Wei, Wang Lei, Liu Yi, et al.Improved rotor flux observer for sensorless control of PMSM with adaptive harmonic elimination and phase com- pensation[J]. CES Transactions on Electrical Machines and Systems, 2019, 3(2): 151-159.
[8] Yin Zhonggang, Gao Fengtao, Zhang Yanqing, et al.A review of nonlinear Kalman filter appling to sensorless control for AC motor drives[J]. CES Transactions on Electrical Machines and Systems, 2019, 3(4): 351-362.
[9] Tu Xuan, Hou Xinguo, Zhao Jinghong, et al.Speed identification of speed sensorless linear induction motor based on MRAS[J]. CES Transactions on Electrical Machines and Systems, 2023, 7(3): 294-300.
[10] Chen Zhiqian, Tomita M, Doki S, et al.An extended electromotive force model for sensorless control of interior permanent-magnet synchronous motors[J]. IEEE Transactions on Industrial Electronics, 2003, 50(2): 288-295.
[11] De Angelo C, Bossio G, Solsona J, et al.Mechanical sensorless speed control of permanent-magnet AC motors driving an unknown load[J]. IEEE Transa- ctions on Industrial Electronics, 2006, 53(2): 406-414.
[12] Solsona J, Valla M I, Muravchik C.On speed and rotor position estimation in permanent-magnet AC drives[J]. IEEE Transactions on Industrial Electronics, 2000, 47(5): 1176-1180.
[13] Zhang Yanping, Yin Zhonggang, Bai Cong, et al.A rotor position and speed estimation method using an improved linear extended state observer for IPMSM sensorless drives[J]. IEEE Transactions on Power Electronics, 2021, 36(12): 14062-14073.
[14] Li Haitao, Zheng Shiqiang, Ren Hongliang.Self- correction of commutation point for high-speed sensorless BLDC motor with low inductance and nonideal back EMF[J]. IEEE Transactions on Power Electronics, 2017, 32(1): 642-651.
[15] Song Xinda, Fang Jiancheng, Han Bangcheng, et al.Adaptive compensation method for high-speed surface PMSM sensorless drives of EMF-based position estimation error[J]. IEEE Transactions on Power Electronics, 2016, 31(2): 1438-1449.
[16] De Kock H W, Kamper M J, Kennel R M. Anisotropy comparison of reluctance and PM synchronous machines for position sensorless control using HF carrier injection[J]. IEEE Transactions on Power Electronics, 2009, 24(8): 1905-1913.
[17] Wei Jiadan, Wang Junjie, Zhang Zhuoran, et al.Frequency-insensitive rotor position estimation method for three-stage synchronous machine based on indirect high-frequency signal injection[J]. IEEE Transactions on Transportation Electrification, 2022, 8(2): 1785-1793.
[18] Han Bangcheng, Shi Yangyang, Song Xinda, et al.Initial rotor position detection method of SPMSM based on new high frequency voltage injection method[J]. IEEE Transactions on Power Electronics, 2019, 34(4): 3553-3562.
[19] Zhang Guoqiang, Wang Gaolin, Yuan Bihe, et al.Active disturbance rejection control strategy for signal injection-based sensorless IPMSM drives[J]. IEEE Transactions on Transportation Electrification, 2018, 4(1): 330-339.
[20] Szalai T, Berger G, Petzoldt J.Stabilizing sensorless control down to zero speed by using the high- frequency current amplitude[J]. IEEE Transactions on Power Electronics, 2014, 29(7): 3646-3656.
[21] Slininger T S, Xu Yinghan, Lorenz R D.Enhancing estimation accuracy by applying cross-correlation image tracking to self-sensing including evaluation on a low saliency ratio machine[C]//2016 IEEE Energy Conversion Congress and Exposition (ECCE), Milwaukee, WI, USA, 2016: 1-7.
[22] Wang Gaolin, Valla M, Solsona J.Position sensorless permanent magnet synchronous machine drives-a review[J]. IEEE Transactions on Industrial Elec- tronics, 2020, 67(7): 5830-5842.
[23] 麦志勤, 肖飞, 刘计龙, 等. 基于改进型自调整轴系幅值收敛电流解调算法的旋转高频电压注入法[J]. 电工技术学报, 2021, 36(10): 2049-2060.
Mai Zhiqin, Xiao Fei, Liu Jilong, et al.Rotating high-frequency voltage injection method based on improved self-adjusting frame amplitude convergence current demodulation algorithm[J]. Transactions of China Electrotechnical Society, 2021, 36(10): 2049-2060.
[24] Raca D, Garcia P, Reigosa D D, et al.Carrier-signal selection for sensorless control of PM synchronous machines at zero and very low speeds[J]. IEEE Transactions on Industry Applications, 2010, 46(1): 167-178.
[25] 吴春, 陈科, 南余荣, 等. 考虑交叉饱和效应的变角度方波电压注入永磁同步电机无位置传感器控制[J]. 电工技术学报, 2020, 35(22): 4678-4687.
Wu Chun, Chen Ke, Nan Yurong, et al.Variable angle square-wave voltage injection for sensorless control of PMSM considering cross-saturation effect[J]. Transactions of China Electrotechnical Society, 2020, 35(22): 4678-4687.
[26] Wang Henghong, Wang Jiyao, Xu Wei.A high precision initial position estimation method for low saliency ratio machine based on image tracking[J]. IEEE Transactions on Power Electronics, 2023, 38(11): 13612-13624.
[27] Kim H, Huh K K, Lorenz R D, et al.A novel method for initial rotor position estimation for IPM synchronous machine drives[J]. IEEE Transactions on Industry Applications, 2004, 40(5): 1369-1378.
[28] Degner M W, Lorenz R D.Using multiple saliencies for the estimation of flux, position, and velocity in AC machines[J]. IEEE Transactions on Industry Applications, 1998, 34(5): 1097-1104.
[29] Martin K.Complex signal processing is not- complex[C]//ESSCIRC 2004-29th European Solid- State Circuits Conference, Estoril, Portugal, 2003: 3-14.
[30] Ganin Y, Lempitsky V.Unsupervised domain adaptation by backpropagation[C]//Proceedings of the 32nd International Conference on Machine Learning, Lille, France, 2015: 1180-1189.