於锋　吴晓新　田朱杰　朱晨光

摘 要：针对永磁同步电机（PMSM）三电平模型预测转矩控制（MPTC）中存在中点电压波动、权值整定困难等问题，提出一种计及中点电压平衡的PMSM三电平无权值模型预测磁链控制（MPFC）方法。首先，将MPTC中对电机转矩和定子磁链幅值的控制转化为对定子磁链矢量的控制，从而构建了仅包含定子磁链dq分量的价值函数，消除了传统MPTC价值函数中的权重系数。在此基础上，根据正负冗余小矢量对中点电位作用效果相反的特性，通过有效筛选冗余小矢量，在不引入额外权值系数的前提下有效解决了三电平逆变器中点电压波动的问题。为进一步提高电机稳态性能，通过矢量分区提出了q轴磁链无差拍的占空比MPFC策略。最后，仿真和实验结果验证了所提方案的可行性和优越性。

关键词：永磁同步电机;三电平;中点电压;权值;模型预测磁链控制;占空比

DOI：10.15938/j.emc.2020.09.016

中图分类号：TM 351

文献标志码：A

文章编号：1007-449X（2020）09-0145-11

Predictive flux control for PMSM without weighting factors using three-level converters with neutral-point voltage balance

YU Feng， WU Xiao-xin， TIAN Zhu-jie， ZHU Chen-guang

（School of Electrical Engineering， Nantong University， Nantong 226019， China）

Abstract：

To address the neutral-point voltage fluctuation and weighting factor tuning work considerations， an improved three-level model predictive torque control strategy （MPTC） was developed in the permanent magnet synchronous motor （PMSM） drive system. Thus， an unweighted three-level model predictive flux control （MPFC） scheme associated with neutral-point voltage balance was proposed. Firstly， the control motor torque and stator flux amplitude in MPTC was converted into the control of stator flux vector， and the cost function associated with only dq components of stator flux was constructed， while the weight factor included in the cost function of traditional MPTC was eliminated. Secondly， the neutral-point voltage fluctuation in three-level inverter is addressed in a manner where the redundant small vector was properly selected according to the characteristics that the negative and positive redundant small vectors have the opposite effect on the neutral point voltage， thus no additional weighting factors are required. Furthermore， to improve the steady-state performance of PMSM， the duty-cycle MPFC strategy based on q-axis flux linkage deadbeat concept was developed. Finally， the simulation and experimental results verify feasibility and superiority of the control scheme.

Keywords：permanent magnet synchronous motor; three-level; neutral point voltage; weighting factor; model predictive flux control; duty cycle

0 引 言

模型預测控制（model predictive control， MPC）产生于1970年后期的工控领域，被认为是继矢量控制和直接转矩控制之后又一种高性能电机控制策略[1]。MPC以系统的数学模型为基础去预测将来的输出状态，并根据价值函数作为优化手段，使得控制灵活性高，具体可分为连续控集MPC[2]和有限控集MPC[3]。不同于连续控集MPC中算法运算量大的缺点，有限控集MPC利用变换器的离散特性，只对逆变器输出的有限个电压矢量进行预测计算，算法更简单易行。近年来，随着数字信号处理器的发展以及MPC本身具有结构简单、动态性能好、可实现多变量优化和非线性约束处理等优点，使得MPC应用到电机驱动领域成为当前研究热点[4-7]。……