模块化多电平矩阵变换器（M3C）作为一种可实现直接交-交电能变换的高压、大容量、模块化电路拓扑，在柔性低频输电和变速抽水蓄能等场合有广阔的应用前景，可帮助解决大量可再生能源接入电网带来的电能并网传输及电网稳定性方面的挑战。本文对M3C在建模分析、电容电压平衡控制、子模块故障容错控制和桥臂故障容错控制方面的若干关键问题进行了深入研究，主要内容包括：首先，针对M3C拓扑分析了其基本电流分布下的电容电压波动特性，并根据M3C中电气量的循环耦合关系分析了电容电压波动造成的桥臂电压波动及其各频率分量在各桥臂中的相位分布，从而建立了M3C环流的一次耦合模型，为使用比例谐振控制器抑制对应频率分量的环流奠定了基础。其次，通过双αβ0变换和对角变换得到了M3C的解耦功率方程，导出了M3C的桥臂间能量平衡解耦控制方法。为解决M3C在输入输出频率接近时存在的电容电压不稳定问题，提出了一种将可用共模电压的最大绝对值实时注入的优化控制方法。相比正弦共模注入法和已有两步优化注入法，所提方法可在环流和桥臂电流幅值更小的条件达到更好的电容电压波动抑制效果。再次，为实现子模块故障情况下的M3C穿越运行，首先研究了M3C在功率器件发生开路故障后的故障检测和容错控制问题，提出了一种对角桥臂组合变换，将环流观测误差变换为特定组合电流的观测误差，基于此得到了一种适用于M3C的功率器件开路故障检测准则，可实现对故障的快速和准确检测。而后，提出了一种自适应最优共模电压注入方法来实现M3C的非冗余子模块故障容错控制，该方法根据桥臂基波参考电压和桥臂最大可用电压实时计算最优共模电压，可在任意条件下实现非冗余子模块故障后M3C输入输出电压范围的最大化。最后，为实现M3C在部分桥臂故障条件下的稳定运行，提出了一种故障后桥臂电流的解析配置方法。所提方法先导出桥臂故障条件下的基本电流配置，再使用解析环流实现桥臂能量平衡，最终得到了单桥臂故障和两桥臂故障条件下的解析桥臂电流配置，导出的解析解对单桥臂故障来说还是最小化最大桥臂峰值电流条件下的最优解。相比已有的离线优化配置方法，所提方法的配置结果可自动适应不同的负载功率因数，无需查表计算；同时可利用所得的桥臂电流解析配置分析桥臂故障后M3C的安全工作区。
The modular multilevel matrix converter (M3C) is a high-voltage, high-capacity, and modular topology that can realize direct AC-AC power conversion, which is promising in several applications such as flexible low-frequency AC transmission and variable-speed pumped-storage power plants. In this dissertation, several key issues of the M3C are deeply studied, including in modeling and analysis, capacitor-voltage balancing control, submodule (SM) fault-tolerant control, and branch fault-tolerant control. The main contents are as follows:First, the capacitor-voltage fluctuation of the M3C under basic branch currents is analyzed. Then, according to the cyclic coupling relationship of electrical quantities in the M3C, the branch voltage fluctuations caused by the capacitor-voltage fluctuations are derived and analyzed. As a result, the first-order coupling model of the circulating currents of the M3C is established, which lays a foundation for suppressing the circulating currents of the corresponding frequency components by employing the proportional resonance controllers.Second, by employing the double αβ0 and diagonal transformations, the decoupled power equations and corresponding decoupled branch energy balancing control method of the M3C are derived. To solve the problem that capacitor voltages are unstable when the input and output frequencies are close, an optimal control strategy is proposed, which suppresses capacitor-voltage fluctuation by always injecting the CMV with the maximum available absolute value. Compared with the sinusoidal CMV injection method and the existing two-step optimal injection method, the proposed method can achieve better suppression results for capacitor-voltage fluctuation with lower circulating currents and branch currents.Third, to realize the ride through of SM faults, the fault detection and tolerant control of the M3C under the open-circuit fault of power semiconductor devices are firstly studied. A diagonal branch-combination transformation is proposed to transform the observation errors of circulating currents into that of specific combined currents. Based on this, an open-circuit fault detection criterion suitable for the M3C is obtained, which can realize rapid and accurate fault detection. Then, an adaptive optimal CMV injection method is proposed to realize the fault-tolerant control under the nonredundant SM fault. This method calculates the optimal CMV in real-time according to the fundamental branch voltage and the maximum available branch voltage, which can maximize the input and output voltage range of the M3C under any condition.Finally, to realize the stable operation of the M3C under branch fault conditions, an analytical branch current configuration method after branch faults is proposed. The proposed method firstly derives the basic branch current configurations under branch fault conditions, and then injects analytical circulating currents to realize the branch-energy balance. As a result, the analytical branch current configurations under one and two branch fault conditions are derived. The derived analytical configuration is also the optimal one for the single branch fault condition. Compared with the existing off-line optimal configuration method, the configuration results of the proposed method can automatically adapt to different load power factors without the requirement of look-up tables. Meanwhile, the derived analytical branch current configurations can be employed to analyze the safe operation area of the M3C under branch fault conditions.
