Search

his research focuses on a three-phase Active Front End (AFE) converter that uses Model Predictive Control (MPC) for its switching operations. The MPC technique forecasts the AFE voltages and currents by integrating the voltage vectors from the converter switches across eight distinct switching states into a discrete state space model. Simulation results reveal a rapid dynamic response to changes in load voltage and minor fluctuations in current and wave values, which increase after connecting to an unbalanced load. Testing the effect of sampling time shows that a 5µs sampling time achieves the lowest Total Harmonic Distortion (THD), with 0.02% for voltage and 0.13% for current. This indicates that smaller sampling times improve the accuracy of reference voltage predictions, thereby enhancing electricity quality. Furthermore, the impact of nonlinear loads on voltage demonstrates that as the load resistance significantly increases from 20 Ω to 45 Ω, inversely proportional to the load inductance value, the THD remains relatively constant, ranging from 0.03% to 0.04%. In the 780th prediction analysis using MPC with a 25µs sampling time, the voltage reference value is determined. The cost function value is 0.0127 in the 8th switching state, with predicted voltage values of Vsa = -51.3829 V, Vsb = -2.53489 V and Vsc = 3.04872 V [ABSTRACT FROM AUTHOR]

Virtual vector is applied in neutral‐point‐clamped (NPC) three‐level converter finite control set model predictive control (FCS‐MPC) strategy due to its good performance of less current ripples and high power efficiency. However, in the process of adding virtual vectors, some redundant virtual vectors will be selected. Excessive virtual vector will increase the computational burden and affect grid‐tied current performance. A virtual vector optimal selection strategy for model predictive control to determine the vectors is proposed. The strategy calculates the fluctuation of grid‐tied current through theoretical analysis. Based on the analysis of amplitude fluctuation of grid‐tied current, the graphics of coverage angle visualization can be obtained under different virtual vector. The angle of the required virtual vectors is obtained through the graphics of coverage angle visualization. The number of virtual vector can be determined when the error current angle range covers the entire vector graph. The proposed control strategy was compared with the conventional control strategies, including excessive virtual vectors or with no any virtual vector. Experimental results show that the proposed control strategy can significantly reduce THD, grid‐tied current error and computational burden. © 2022 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]