چكيده به لاتين
Modular Multilevel Converters (MMC) are well-developed options for various applications such as HVDC systems and electrical drives, due to their several advantages like improved output power quality, reduced voltage stress on the power semiconductors, common mode voltage decrease, modular structure, and etc. . However, one on the main issues related to this topology, is its complicated control structure. The reason behind this is existence of various control aspects such as output current control, elimination of circulating current control, and balancing of arm and also cell capacitor voltages. In this thesis, a novel control scheme is proposed for MMC based on the Model Predictive Control (MPC) concept, which not only solves the drawbacks of the conventional control approaches, but also notably improves the MPC method regarding its huge computational burden. The proposed method is developed based on the deadbeat approach principles. In the deadbeat approach, the reference voltage of the inverter is determined at every instance in a way that the difference between the future control variable and its reference value tends to become zero. The calculation of the reference voltage is conducted based on the discrete system model. In the proposed method in this work, output current control and elimination of the circulating current are considered as the primary control goals and based on these, the reference voltages of upper and lower arms of each phase of MMC is calculated. Next, by conducting some simple calculation, the approximate voltage levels of the upper and lower arms are determined based on those voltage references. After this, since there is notable free time for conducting calculations, the MPC approach is performed in order to determine the exact voltage levels. It should be noted that only four different combinations of the upper and lower voltage levels are examined for each phase and this value will be 12 for the three phase case. The main advantage of the proposed method is the independency of the examined states from the cell numbers. Finally, a voltage balancing stage is performed in order to determine the switching states based on the optimum voltage levels. In the last part of the report, different aspects of the proposed method are verified by conducting simulations in the MATLAB/Simulink environment.
