Improved double-mode model predictive control scheme with reduced output dv/dt for modular multilevel converters

Despite several advantages of modular multilevel converters (MMCs), like easy scalability, modularity, and better output quality, it is required to control their internal dynamics, besides the output current control. This paper proposes a double-mode model predictive control (MPC) scheme for three-phase MMCs, with highly-reduced computational load. Because of MPC inherent features like the capability of including several control objectives in a single cost function and excellent dynamic performance, it is considered as an attractive choice for controlling MMCs. However, the huge computational burden is an important obstacle to implement MPC for this topology. The proposed scheme consists of two separate modes for the dynamic and steady-state conditions, in which the admissible candidate vectors are appropriately restricted, based on the special requirements of each operating state. In the dynamic operating mode, achieving fast response is significantly important; while circulating current reduction and lower output dv/dt are considered as main control indexes at the steady state. Moreover, different exclusive cost functions are utilized in each mode, which would result in further reduction of the prediction calculations. Finally, the simulation and experimental results verify the satisfactory steady-state and dynamic performances, with balanced submodule capacitor voltages and reduced computational burden.