Simplified MPC Algorithm of NPC Three-level Inverter Based on SVPWM

The traditional control method of the NPC （neural point clamped） three-level inverter is unable to effectively reduce CMV （common-mode voltage） and needs a large calculation quantity because of the zero vector involved modulation or rolling optimization. Therefore， this paper proposes a simplified MPC （model predictive control） based on SVPWM （space vector pulse width modulation）. Firstly， the normalized control of output voltage is conducted， and the sector is determined by symbol judgment and logical operation. The 27 vectors selected in a single cycle are reduced to 10 vectors， which resolves the calculation complexity. Then a multi-objective optimization function including midpoint potential deviation， CMV suppression and switching frequency is constructed， and the vector which is best for the inverter is selected to minimize CMV and keep the switching frequency constant. Finally， it is verified by an experiment that the method can control multiple targets in a shorter time in the condition of the same ha