基于高斯过程回归的永磁同步电机非线性磁链辨识  

Nonlinear Magnetic Flux Identification of Permanent Magnet Synchronous Motors Based on Gaussian Process Regression

作　　者：刘忠永 范涛 何国林[2] 温旭辉[1,2] LIU Zhongyong;FAN Tao;HE Guolin;WEN Xuhui(University of Chinese Academy of Sciences,Beijing 100049,China;Institute of Electrical Engineering,Chinese Academy of Sciences,Beijing 100190,China)

机构地区：[1]中国科学院大学,北京100049 [2]中国科学院电工研究所,北京100190

出　　处：《电源学报》2024年第3期172-181,共10页Journal of Power Supply

基　　金：国家重点研发计划资助项目(2021YFB2500600)。

摘　　要：在新能源汽车发展领域,以碳化硅为代表的新一代半导体功率器件正在逐步取代硅基IGBT,崭新的技术生态对电机控制性能也有了更高的要求。从传统的PI控制、直接转矩控制到模型预测控制、神经网络控制等新算法,电机参数的精准度逐渐成为电控系统进一步提升性能的关键因素。针对永磁同步电机经典线性模型受交叉饱和等非线性因素影响不能适用于复杂多变工况的问题,提出基于高斯过程回归的非线性磁链辨识方法,使用二阶广义积分器获取动态工况中的磁链数据完成系统辨识,通过仿真与实验验证了该方案的有效性及参数辨识的准确性。In the burgeoning field of new energy vehicles,silicon carbide representing a new generation of semiconductor power devices is progressively replacing silicon-based IGBTs,which also sets higher standards for the motor control performance within the corresponding innovative technological ecosystem.The precision of motor parameters is becoming increasingly critical for enhancing the performance of electric control systems as they evolve from the traditional PI control and direct torque control to advanced algorithms such as model predictive control and neural network control.Aimed at the problem that the classic linear model for permanent magnet synchronous motors cannot adapt to complex and variable conditions due to nonlinear factors such as cross-saturation,a nonlinear magnetic flux identification method based on Gaussian process regression is proposed.By employing a second-order generalized integrator to acquire the magnetic flux data under dynamic conditions,the system identification is completed.Finally,the effectiveness of the proposed approach and the accuracy of parameter identification were verified through simulation and experimental results.

关 键 词：碳化硅 电机控制 参数辨识 高斯过程回归 

分 类 号：TM351[电气工程—电机]