Variable Switching Point Model Predictive Control for DC-Link Voltage Regulation of Back-to-Back Converters

Abstract

In this article, a novel control method for back-to-back converters used in grid-to-motor connections is explored. To increase the robustness of low dc-link capacitances, a control method based on variable switching point model predictive control is proposed. While previous model predictive control methods for the back-to-back converter selected a certain switching state to fulfill all control goals, we use the switching time in addition to the switching state in order to minimize deviations from the target voltage. Choosing a variable switching point provides an additional degree of freedom to the control framework and allows the system to cope with the large number of control variables. In this case, the variable switching point is used to minimize the effects of low dc-link capacitances on the system. This can either be achieved by selecting a switching point that yields low dc-link capacitor charging or by selecting a switching point that aims to keep the dc-link voltage close to the reference. The proposed method is verified through numerical simulations and hardware-in-the-loop experiments and compared to existing approaches. The results show that it is possible to control the dc-link using only the switching point of the converter.