A Predictive Controller for Modular Multi-level Converters

Abstract

This paper presents a novel predictive controller for modular multi-level converters (M2LCs). Different from existing model predictive control (MPC) strategies, the control variables are separated into two groups with different control schemes. The load current, controlled through the model predictive direct slope control (MPDSC), is regulated within symmetrical bounds around the reference. And the switching frequency, circulating current and capacitor voltage ripples are minimized by a model predictive multilayer control (MPMC), where each variable is evaluated independently on their own control layer. An advantage of the proposed strategy is to allow operation at low switching frequency by forcing switch transitions to take place only near the bounds and selecting a unique output voltage level with the minimum slope of the load current, relative to the reference. Moreover, a hierarchical screening algorithm is proposed to reduce the number of switching states to be evaluated for each control variable. And a further reduction in the overall computational burden can be achieved by assigning variables to their appropriate control layers. Besides, a novel method of tuning weighting factors is proposed to tune the trade-off among different control variables, which significantly improves the efficiency of the tuning procedure. Simulation results of a single-phase, three-level, 380-VA M2LC, controlled through the proposed strategy under both steady-state and transient conditions are given to demonstrate the validity of the proposed strategy.