Research On Three-level NPC Inverter Base On DSPACE

With the increasing energy crisis, the application of power electronics technology in the smart grid, new energy power generation and rail transport is increasing. The three-level converter in high voltage high power applications has attracted more and more attention. This paper focuses on the control technology of three-level NPC inverter and the interconnection between inverter and the high impedance grid based on dSPACE simulation system.This paper analyses the work principle of three-level inverter, establish its mathematical model and design the corresponding controller, then introduces two modulation algorithms of three-level space vector. Then the neutral point balance problem and the reason for the neutral point voltage fluctuation are analyzed in detail. This paper also analyzes the relationship between control parameters and the neutral point voltage fluctuation on the ideal situation for hysterics control scheme. And get the appropriate parameters by comparing the neutral point voltage size and balancing control ability under different parameters. Then this paper establish mathematical model of the neutral point voltage, describes the reason of the control effect of active control will be affected by working conditions of inverter and solve this problem by introducing the reciprocal of the neutral point current which caused the small vector. Resonant controller is applied and the neutral point current which caused the medium vector is introduced for feed forward control to eliminate low-frequency voltage ripples. The above analysis and active controller design method is verified by experiment.This paper also analyzes the influence of grid detection point voltage of grid-tie inverter for high-impedance grid. In order to solve zero crossing distortion of detection voltage, this paper uses high frequency capacitance voltage compensation to inhibiting LCL resonant, with repetitive control to eliminate the impact of dead zone. The experiments verify the effectiveness of the measures.