A Direct Model Predictive Control Strategy of Back-to-Back Modular Multilevel Converters Using Arm Energy Estimation

This paper presents a model predictive control (MPC) algorithm for modular multilevel converters (MMCs). To meet the control objectives of phase current reference tracking and circulating current minimization, the proposed control scheme calculates the optimal number of submodules (SMs) to be inserted in each arm. In doing so, favorable steady-state and dynamic performance is achieved. More-over, by estimating-instead of measuring-the arm energies in the predictive stage of the control loop, the proposed control scheme results in self-stabilizing open-loop arm energy balancing, while avoiding potential stability issues. Furthermore, to reduce the computational complexity of the MPC algorithm, the optimization problem is simplified by controlling each phase separately and assuming that the SM capacitors are balanced within an arm. To ensure that this assumption is always satisfied, a subsequent capacitor voltage balancing algorithm is designed to select the individual SMs that are switched on and off. The performance of the proposed control strategy is validated with simulations for a high voltage dc system (HVDC) that consists of two MMCs with 20 SMs per arm in a back-to-back configuration.