基于多参数辨识的永磁同步电机磁链预测控制  

Multi-Parameter Identification Based Flux Predictive Control for Permanent Magnet Synchronous Motor

作　　者：王祥松 杨淑英[1] 谢震[1] 张兴[1] Wang Xiangsong;Yang Shuying;Xie Zhen;Zhang Xing(School of Electrical Engineering and Automation Hefei University of Technology,Hefei,230009,China)

机构地区：[1]合肥工业大学电气与自动化工程学院,合肥230009

出　　处：《电工技术学报》2024年第22期7084-7098,共15页Transactions of China Electrotechnical Society

基　　金：国家自然科学基金资助项目(51877062)。

摘　　要：在永磁同步电机(PMSM)有限集模型预测磁链控制策略(FCS-MPFC)中,磁链给定值计算、实际值获取以及未来值预测都要用到电机参数。受工况和温度的影响,电机参数具有变化性,参数偏差将影响实际预测控制性能。为提高控制系统鲁棒性,需要对电机d、q轴电感和永磁磁链进行在线辨识。鉴于此,该文提出了永磁同步电机多参数辨识方案,并基于此设计了FCS-MPFC策略。首先,依据电机稳态方程,设计了广义比例积分观测器(GPIO)对PMSM的q轴电感和永磁磁链进行辨识。其次,利用FCS-MPFC中电机定子电流变化较大的特征,提出了基于电流变化量的d轴电感辨识方法,克服了多参数辨识的欠秩问题。再次,依据FCS-MPFC策略下逆变器电压输出特性,优化设计了电流采样,有效避免了逆变器死区对辨识效果的影响。最后,针对磁链预测过程,设计了基于GPIO的磁链预测模型,进一步提高磁链预测的参数鲁棒性。该文的分析、设计以及性能提升均得到实验验证。In finite control set-model predictive flux control(FCS-MPFC)for permanent magnet synchronous machine(PMSM),the parameters are required for flux linkage reference calculation,actual flux estimation,and flux prediction in the following one or two sampling periods.Due to the influence of operation states and temperature,machine parameters are ready to deviate.Therefore,identifying their values online is crucial to the performance of the actual predictive control system,including the d-and q-axis inductances and the permanent magnet flux.This paper proposes a multi-parameter identification scheme for PMSM.Consequently,with the identified parameters,the flux is treated,and an FCS-MPFC scheme is implemented.Firstly,this paper designs a generalized proportional integral observer(GPIO)to observe the disturbance caused by parameter deviations.The motor parameter deviations in the disturbance can be extracted based on the mathematical model of the motor.Then,the q-axis inductance and permanent magnet flux can be accurately identified by compensating the nominal parameters.Next,a d-axis inductance identification method is proposed based on the stator current variation characteristics under the FCS-MPFC strategy,which constructs an identification equation for the d-axis inductance using the d-axis current difference equation.It improves identification accuracy by setting appropriate boundary conditions for the differential current values,overcoming the rank deficiency problem in multi-parameter identification.In addition,boundary conditions are set for current and speed to ensure the identification effectiveness.The low-pass filters eliminate high harmonic noises in the identified motor parameters.Furthermore,the identification process is improved by leveraging the characteristics of inverter voltage output under the FCS-MPFC strategy.Specifically,two current samples are taken within a control cycle.By appropriately setting the sampling moments,the dead zone voltage is eliminated from the identification equation.Finally,for

关 键 词：有限集模型预测磁链控制 广义比例积分观测器 多参数辨识 参数鲁棒性 

分 类 号：TM351[电气工程—电机]