NPC型三电平逆变器可视化三矢量无模型预测控制策略

为了解决传统模型预测控制参数鲁棒性差、输出电流纹波大的不足，提出了一种应用于三相三电平逆变器的可视化三矢量无模型预测控制策略。该方法首先利用滑模观测器估计系统集总参数，采用无差拍控制获得参考电压矢量。该过程无需任何系统参数，提高了控制策略的参数鲁棒性。其次，在每个采样周期内施加由冗余矢量构成的三矢量组合，减少了输出电流纹波并实现了电容电压平衡。在此基础上，提出了一种新的可视化分析方法，从理论上证明了所提出的简化三矢量寻优的有效性。实验结果表明，在模型参数失配的情况下，所提策略可使NPC三电平逆变器的并网电流拥有较低的电流总谐波失真率，增强了模型预测控制的参数鲁棒性。

Visualized three-vector model-free predictive control strategy for an NPC three-level inverter

To overcome the drawbacks of poor robustness of traditional model predictive control parameters and large output current ripple, a visual three-vector model-free predictive control strategy for three-phase three-level inverters is presented. First, the sliding mode observer is used to estimate the lumped parameters of the system, and the reference voltage vector is obtained using deadbeat control. This process does not require any system parameters and improves the robustness of the control strategy parameters. Second, a three-vector combination consisting of redundant vectors is applied during each sampling period to reduce the output current ripple and achieve capacitive voltage balancing. A new visual analysis method is presented, and the effectiveness of the simplified three-vector optimization method is proved in theory. The experimental results show that the proposed strategy can make the grid-connected current of NPC three-level inverters have a lower total harmonic distortion rate when the model parameters are mismatched. This enhances the parameter robustness of the model predictive control.