六相永磁同步发电机(PMSG)谐波含量较低、解耦方法成熟且稳定性强，具有广阔的应用前景。为提升六相PMSG负载转矩突变时的动态性能，增强系统的抗干扰能力，提出了一种利用负载转矩观测值前馈补偿的策略。首先，采用滑模观测器观测负载转矩，观测器滑模面设计为实际转速与观测转速之差，在观测过程中滤除了抖振信号，且未引起相位和幅值变化；其次，将负载转矩观测值按比例前馈至电流环给定值，弥补滑模控制器不连续项幅值不足的问题；再次，针对前馈比例常数过大导致转速无法快速稳定在给定值的问题，将前馈比例系数设为可变值;最后，基于Matlab/Simulink建立了多相模块化多电平换流器（MP-MMC）驱动六相PMSG负载转矩观测前馈滑模控制的仿真模型，对所提策略进行验证。仿真结果表明滑模观测器可准确观测负载转矩，前馈补偿策略有效抑制了负载突变引起的转速波动。

Six-phase permanent magnet synchronous generator (PMSG) has broad application prospects due to its low harmonic content, mature decoupling methods, and strong stability. To improve the dynamic performance of the six-phase PMSG when the load torque suddenly changes, and to enhance the system’s anti-interference ability, a feedforward compensation strategy using the load torque observation value is proposed. Firstly, the load torque is observed by the sliding-mode observer, the sliding surface of the observer is designed as the difference between the actual speed and the observed speed, and the chattering signal can be filtered out during the observation process without causing phase and amplitude changes. Secondly, the load torque observation value is proportionally feed forward to the given value of the current loop, making up for the problem of insufficient amplitude of the discontinuity term of the sliding mode controller. Thirdly, for the problem that the feedforward proportion constant is too large, resulting in the speed not being able to quickly stabilized at the given value, the feedforward proportion coefficient is set to a variable value, which effectively suppresses the speed fluctuation. Finally, based on Matlab/Simulink, the simulation model of multi-phase-modular multilevel converter (MP-MMC) drive six-phase PMSG load torque observation feedforward sliding-mode control is established to validate the proposed stratage. The simulation results show that the sliding-mode observer can accurately observe the load torque, and the feedforward compensation strategy effectively suppresses the speed fluctuation caused by the sudden load change. 

国家自然科学基金 (52077135); 国网上海市电力公司科技项目（5209KZ220002）