Cost function‐based modulation scheme of model predictive control for VIENNA rectifier

Abstract

Finite control set model predictive control (FCS-MPC) has been widely used in the control of grid-connected converters with the advantages of fast dynamics, multi-objective control, and easy implement. However, the conventional FCS-MPC bears with variable switching frequency, high current ripple and computational burden. An improved current model predictive with the cost function-based modulation scheme (CFM-MPC) is proposed for a three-phase three-level VIENNA rectifier to improve the power quality. First, the mathematical model and voltage vector are given according to the principle of deadbeat control. Then, the voltage vector of different voltage vectors are selected according to the location of the voltage vector reference, and the switching action time of the selected are directly calculated by the inversely proportional with cost function value of the selected vectors. It remains the merits of both the conventional MPC and space vector pulse width modulation schemes to track the optimum voltage vector without increasing the computational burden. Finally, a comparative study with the proposed CFM-MPC and conventional FCS-MPC has been conducted to verify the superiority of the proposed scheme. The results show the proposed CFM-MPC has the advantages of lower power ripple, fixed switching frequency, lower total harmonic distortion and neutral point potential balance.

Model predictive duty cycle control for three‐phase Vienna rectifiers

“…However, the current ripple is relatively large because the optimal voltage vector is applied during the whole sampling period. Aiming to reduce the current ripple and improve the steady-state performance of the FCS-MPC method without increasing the sampling frequency, some multi-vector MPC strategies have been reported [23][24][25][26][27]. A discrete space vector modulation MPC method is presented in [23].…”

“…Given that only two real voltage vectors are utilized to synthesize the virtual voltage vectors for the grid current control, the steady-state performance is still not satisfactory. To further enhance the steady-state performance, three or four voltage vectors are applied in every sampling period in [24][25][26][27]. In [24], three real voltage vectors are utilized to synthesize the virtual voltage vectors.…”

“…In [24], three real voltage vectors are utilized to synthesize the virtual voltage vectors. In [25], four voltage vectors are selected based on the location of the grid voltage vector, and the duty cycles for different voltage vectors are set according to the values of their cost functions. In [26], three voltage vectors are selected by pre-selecting one redundant vector, and their duty cycles are obtained by minimizing the cost function using a least-square optimization method.…”

“…To reduce the computational burden of the method in [26], the equivalence relation between different switching sequences is derived, and a computationally efficient optimal switching sequence model predictive control (CE-OSS-MPC) is proposed in [27]. However, these methods [24][25][26][27] are directly optimizing the switching sequences. Thus, they require to store the switching states of all the voltage vectors, which increases the memory usage.…”

Model predictive direct power control scheme for Vienna rectifier with constant switching frequency

“…In [15], the direct torque control for Vienna rectifier considers not only the torque and flux, but also the neutral-point voltage and then the optimized voltage vector generates by the lookup table. In [16], a cost function based model predictive control modulation scheme is proposed. It can output three voltage vectors in each switching cycle, which effectively reduces the grid current total harmonic distortion (THD) and achieve a fixed switching frequency.…”