基于DUKF参数辨识的永磁电机无差拍预测控制  

Deadbeat Predictive Control of Permanent Magnet Motor Using DUKF Parameter Identification

作　　者：徐玺声 颜黎明 郭鑫 魏巍 孙柯柯 XU Xisheng;YAN Liming;GUO Xin;WEI Wei;SUN Keke(School of Automobile,Chang’an University,Xi’an 710064)

机构地区：[1]长安大学汽车学院,西安710064

出　　处：《电气工程学报》2025年第1期65-77,共13页Journal of Electrical Engineering

基　　金：国家自然科学基金(52107036);陕西省创新能力支撑计划(2021TD-28);陕西省自然科学基础研究计划(2021JQ-252);中央高校基本科研业务费(自然科学类)(300102223201)资助项目。

摘　　要：永磁同步电机(Permanent magnet synchronous motor,PMSM)在运行过程中由于磁路饱和及温升等原因导致电机参数实时变化,控制器参数失配将引起无差拍预测控制性能下降,甚至造成电机驱动系统失稳。针对此问题,提出基于双无迹卡尔曼滤波器(Dual unscented Kalman filter,DUKF)的永磁同步电机无差拍预测控制,DUKF实时估计电机参数并更新无差拍预测模型,以减小参数误差对无差拍控制性能的影响。详细地阐述了DUKF参数辨识方法的基本原理、估计器构造及参数设计方法。同时,对比研究双扩展卡尔曼滤波器(Dual extended Kalman filter,DEKF)、无迹卡尔曼滤波器(Unscented Kalman filter,UKF)及扩展卡尔曼滤波器(ExtendedKalmanfilter,EKF)等参数辨识方法。在试验部分,系统性地研究DUKF、DEKF、UKF及EKF等四种方法在电机稳态、动态下的参数估计误差及收敛速度,分析了四种方法的初值敏感性和算法复杂性,评估了融合DUKF的永磁同步电机无差拍预测控制性能。四种参数辨识方法的性能评估试验结果表明,DUKF的参数估计精度优于其他三种方法。Permanent magnet synchronous motor(PMSM)undergoes real-time variations in motor parameters during operation due to factors such as magnetic saturation and temperature rise.Parameter mismatch in the controller can cause a decrease in the control performance of deadbeat predictive control and even lead to instability in the motor drive system.To address this issue,a PMSM deadbeat predictive control using the dual unscented Kalman filter(DUKF)is proposed.DUKF enables real-time estimation of motor parameters and updates the deadbeat predictive model to mitigate the impact of parameter errors on control performance.The basic principles,estimator construction,and parameter design methods of DUKF are elaborated in detail.Additionally,a comparative study is conducted on parameter identification methods using the dual extended Kalman filter(DEKF),unscented Kalman filter(UKF),and extended Kalman filter(EKF).In the experimental section,the parameter estimation errors and convergence speeds of DUKF,DEKF,UKF,and EKF methods under motor steady-state and dynamic conditions are systematically studied.The sensitivity to initial values and algorithm complexity of four methods are analyzed.Eventually,the performance of integrating DUKF in deadbeat predictive control of PMSM is evaluated.The experimental results of performance evaluation for the four parameter identification methods demonstrate that the parameter estimation accuracy of DUKF is superior to the other three methods.

关 键 词：永磁同步电机 参数估计 双无迹卡尔曼滤波器 双扩展卡尔曼滤波器 无差拍控制 

分 类 号：TM341[电气工程—电机]