Modulated Model Predictive Control for Modular Multilevel AC/AC Converter

Abstract

Modular multilevel converter (MMC) is very popular in high power applications due to its attractive characteristics. The full bridge-based MMC featuring three-phase to single-phase direct ac/ac conversion is a potential solution for ac power supply (ACPS). In this paper, new modulated model predictive control (MMPC) methods are proposed for MMC-ACPS to improve the steady state multi-objective current tracking performance. In the proposed methods, modulated vectors sequence is employed, output voltage levels of upper and lower arms are combined and represented by vectors in the current increments plane first. Then, the plane is divided into eight sectors according to the predictive tracking errors of input current and circulating current. After determining sectors, a modulated vector sequence, consisting of one zero vector and two active vectors, is selected to eliminate these current tracking errors simultaneously at the end of each control period. Duty cycles of the three selected vectors are calculated based on the principle of multiple current tracking errors minimization. Optimized vector selection approaches are illustrated in detail for the proposed MMPC. Since only the adjacent nine vectors are utilized in the proposed methods, the calculation amount is suitable and the dv/dt of the output voltage is also limited. Finally, steady-state and dynamic performances of the proposed control methods are verified by experimental results.