Optimal controller design for back-to-back modular multilevel converter for HVDC systems

Model predictive control (MPC) has gained a lot of attention for power electronic systems during the last two decades. Its ability to control multiple-input multiple-output (MIMO) and non-linear systems, apply constraints on the state, input and output variables, compensate for time delays, and achieve multiple, often conflicting, control objectives with a single control loop makes it well suited for power electronic systems. Challenges in implementing direct MPC in power electronic systems relate to its profound computational burden to solve the optimization problem in real time within a few tens of microseconds. A promising converter topology for medium to high power applications is the modular multilevel converter (MMC). Due to the MIMO nature of the MMC, MPC is a good choice to control the MMC. The possible switching states of the MMC increase exponentially with the number of submodules per MMC arm which further increases the computational burden of the controller. In this thesis, a direct MPC scheme with reference tracking for the MMC is presented and it is validated by simulations for a back-to-back MMC-based high-voltage direct current (HVDC) system with 20 submodules per arm. The controller is capable of achieving low output current total demand distortion (TDD), eliminating the circulating currents and balancing the submodule capacitors while operating at low switching frequency per switching device