为了降低感应电机传统模型预测控制(MPC)因电压矢量作用时间固定引起的转矩和磁链脉动，采用转矩无差拍(DB)控制优化电压矢量作用时间，并从精简电压矢量个数和消除权重系数对传统转矩DB控制进行优化。通过数字仿真和单片机试验对感应电机传统MPC、转矩无差拍模型预测控制(DB-MPC)和提出的优化策略的控制性能和实时性进行对比。仿真结果表明：传统MPC可通过扩充电压矢量来降低转矩脉动。DB-MPC可显著减小转矩脉动、磁链脉动和定子电流谐波含量(THD)，可通过扩充电压矢量来减少磁链脉动和电流THD。优化控制策略减少备选电压矢量，无需权重系数，控制性能与DB-MPC基本相当。基于STM32F103单片机的计算耗时试验表明增加备选电压矢量显著增加计算耗时。相同备选电压矢量下，传统MPC与DB-MPC计算耗时基本相当。在相同控制性能下，优化控制策略可显著减少计算耗时，提高实时性。

In order to reduce ripple of torque and stator flux in conventional model predictive control (MPC) for induction motor due to the fixed action time of voltage vectors, deadbeat (DB) control is used to optimize the action time. An optimized strategy is proposed to simplify candidate voltage vectors and eliminate weighting factor. Control performances and real-time performances of the conventional MPC, the conventional deadbeat model predictive control (DB-MPC) and the proposed strategy are compared through simulation and experiment. Simulation results show that conventional MPC can decrease torque ripple by extending voltage vectors. Compared with MPC, DB-MPC can decrease torque ripple, stator flux ripple and total harmonic distortion (THD) of stator current dramatically and it can decrease stator flux ripple and THD of current by extending voltage vectors. The proposed strategy decreases candidate voltage vectors, eliminates weighting factor and keeps the control performances almost the same as the conventional DB-MPC. Time consumption results based on STM32F103 chip show that extending candidate voltage vectors will increase calculation burden. With the same candidate voltage vectors, the calculation burden of conventional MPC is almost the same as the conventional DB-MPC. With the same control performances, the proposed strategy can decrease time consumption markedly and improve the real-time performance.

陕西省自然科学基金项目(2021JM-163)