An Advanced Model Predictive Control Method Based on Discrete Space Vector Modulation for Reduced Switch Counts Three-Level Converter

The existing model predictive control (MPC) method effectively suppresses common mode voltage (CMV) of the reduced switch counts three-level converter (RSC-TLC), but it cannot properly deal with the problem of neutral point (NP) voltage balance, fixed switching frequency (FSF) and low current ripples simultaneously. To overcome these obstacles, in this article, an advanced MPC method based on discrete space vector modulation (DSVM) for the RSC-TLC is first proposed. First, to reduce the CMV of the RSC-TLC, the CMV amplitudes of all acting voltage vectors are limited to be less than one sixth of dc-link voltage. Second, based on the switching characteristics of the RSC-TLC and the principle of DSVM, numerous virtual vectors are introduced and served as candidates to track the reference current. Third, considering NP voltage balance, the proper switching sequences with adjustable switching time are adopted to output the optimal virtual vector, which has the smallest current tracking error. Consequently, NP voltage balance and FSF with reduced current ripples are achieved to improve current quality. Further-more, only three candidate vectors are evaluated in each predictive step, which avoids evaluating extensive virtual vectors. Meanwhile, the coordinates of candidate vectors and the switching time for both coupled module and independent module are obtained by simple algebraic calculations under gh coordinate system, so that complex trigonometric functions are avoided. As a result, the proposed MPC method is computationally efficient. Finally, the simulation and experimental results verified feasibility and correctness of the proposed MPC method. Remarkably, for the proposed MPC method, the current harmonics are less than 4%, the CMV amplitude is reduced by 50% without significant increase of calculation burden compared with the conventional MPC method.