模块化多电平变换器改进两段式模型预测控制

基于有限控制集的模型预测控制应用于模块化多电平变换器(MMC)时通常会暴露出计算复杂与交流侧输出电流严重失真等问题。提出一种双矢量优化谐波电流模型预测控制策略，即改进两段式模型预测控制，可减少开关状态组合计算量，进一步优化交流侧输出电流。推导了MMC离散数学模型，并构建无权重因子的代价函数，通过计算实际电流与参考电流包围的最小面积重新定义同一周期内两阶段开关信号的占空比，并引入一种附加冗余子模块的方法抑制桥臂环流，接着提出一种败者树多路归并排序算法均衡电容电压。在MATLAB/Simulink平台搭建三相23电平MMC仿真模型验证该策略的有效性。相比于两段式模型预测控制，所提策略的交流侧输出电流跟踪更加精准，谐波畸变率更小。

Improved two-stage model predictive control for modular multilevel converters

The application of model predictive control(MPC) based on finite control set(FCS) to modular multilevel converter(MMC) usually exposes problems such as complex calculation and serious distortion of output current at AC side. A dual-vector optimization harmonic current model predictive control strategy is proposed, which is an improved two-stage model predictive control, and can reduce the calculation amount of switch state combination and further optimize the output current at AC side. The MMC discrete mathematical model is derived, and a cost function without weight factors is constructed, and the duty cycle of the two-stage switching signal in the same period is redefined by calculating the minimum area surrounded by the actual current and the reference current, and an additional redundant sub-module method is introduced to eliminate the bridge arm circulating current, and then a loser tree multi-way merge sort algorithm to balance the capacitor voltage is proposed. A three-phase 23-level MMC simulation model is built on the MATLAB/Simulink platform to verify the effectiveness of the strategy. Compared with the two-stage model predictive control, the proposed control strategy is more accurate in tracking the output current of AC side and less in harmonic distortion rate.