Modeling and Control of a Solid-State Transformer with Small Passive Components for DC Grid Applications

The modern electrical power grid increasingly features semiconductor-based power electronics converters in place of traditional transformers. Known as solid-state transformers (SST), these converters are distinguished for their controllability, their efficiency, and their reduced size and weight. Of the existing converter topologies, the modular multilevel converter (MMC) is popular for implementation as a solid-state transformer, including in direct current (DC) grid interfaces. In the existing engineering literature, the MMC is often studied operating under a low-frequency sinusoidal modulation, and is often implemented with large passive components. This tends to make the converter more stable, and relaxes the requirements placed on the control system. By changing the MMC so that it instead operates with a medium-frequency square waveform in a DC SST, the passive components can be significantly reduced, as well as the semiconductor device ratings. However, a greater emphasis and burden is then placed on the control system’s speed and accuracy. This is problematic because an MMC with small passive components experiences tight coupling among its state variables, which significantly complicates efforts to design a traditional control system for it. This dissertation proposes and investigates methods for modeling and controlling an MMC that uses small passive components. Two major strategies are investigated: two approaches based on the traditional proportional-integral (PI) control, and one approach based on the increasingly popular model predictive control (MPC). By taking advantage of the square-wave operation, it is possible to design simplified model-based PI controllers that achieve power flow control. A more complete model that fully accounts for the MMC’s internal dynamics is also proposed and studied in this dissertation. Although it is difficult to apply the full benefits of this model to PI controller design, an MPC-type controller can easily make full use of it and achieve good control of the MMC’s power flow and internal dynamics. Unlike most other MPC controllers in the engineering literature, the MPC controller proposed in this dissertation has a very low computational burden and does not require iterative solutions. Simulation and experimental results are provided to validate the proposed models and the developed control systems.