Control and fault tolerance improvement of the class of modular multilevel converters

Abstract

The family of the modular multilevel converters (MMCs) has become the most promising class of converters for medium/high-power energy conversion systems, because of salient features including modularity, scalability, high efficiency, superior harmonic performance, easy realization of the redundancy, and absence of DC-link capacitors. The salient features of the MMCs can potentially be exploited for various applications at medium- and high-voltage levels including adjustable speed drives, solid-state transformers, grid integration of renewable energy resources, etc. However, for each application, unique operational and control challenges are imposed. As a result, unique control strategies and design considerations are required to guarantee proper operation of the MMCs, achieve high power density, efficiency, and reliability. Over the past decade, significant research has been performed to address the technical challenges associated with control and operation of various MMC topologies and to unveil their full promises. This thesis is in continuation of this endeavor. On the control aspect, a sliding mode control-based method is proposed for the DCAC MMC. Based on the analysis of system dynamics under the proposed control method, relations among the control parameters and their validity conditions are obtained, which provide a guidance for the systematic controller design. In addition, an active thermal loading control method is also proposed, where a multi-objective optimization problem is formulated and solved to establish a trade-off between the total power losses and the power loss of the semiconductor devices with the highest thermal loading. The Pareto optimal solutions determine the optimal values of the circulating currents, output common-mode voltage, and capacitor voltage reference. On the reliability and fault-tolerance front, a clustering algorithm-based fault detection and locating method is proposed to detect the abnormal curve characteristics of any faulty submodule (SM) of a DC-AC MMC and pinpoint the faulty SM(s). Without any additional sensor, the proposed method can accurately detect and locate the faulty SMs within a reasonable time. Subsequent to bypassing the faulty SMs, the maximum attainable terminal voltage of the faulty phase of the DC-AC MMC is reduced. To maximize the available output line-to-line voltages under the condition, a fault-tolerant control method based on adaption of the neutral-shift strategy and the supporting circulating current suppression method are proposed. For an AC-AC MMC, fault-tolerant control methods for external line-to-ground faults as well as internal SM faults are proposed. Under any external line-to-ground fault, the input-side current regulation, capacitor voltage balancing, and capacitor voltage fluctuation mitigation methods are redesigned to countermeasure the adverse impact of the negative-sequence voltage components in the input side. To address the internal SM fault, the SM capacitor voltages of the faulty cluster are modified to keep the voltage output capability of the faulty cluster. To address the technical challenges associated with low-frequency operation of the DC-AC MMC, a hybrid MMC-derived topology along with its supporting control strategy are proposed to suppress the large SMcapacitor voltage fluctuations without injecting any loadside common-mode voltage. In addition, an AC-AC MMC-based topology is developed to eliminate the large capacitor voltage ripple issue of the conventional AC-AC MMC and to enable the AC-AC MMC for partially-rated solid-state transformers.