Quick Search Algorithm-Based Direct Model Predictive Control of Grid-Connected 289-Level Multilevel Inverter

Abstract

Multilevel inverters, known for their low switching loss and suitability for medium- to high-power applications, often create a heavy computational overhead for the controller. This paper addresses the aforementioned limitation by presenting a novel approach to Direct Model Predictive Control (DMPC) for a grid-tied 289-level ladder multilevel inverter (LMLI). The primary objective is to achieve perfect inverter current control without enumeration. The proposed control method provides a single best solution without complete exploration of the search space. This generalized method can be applied to any multilevel inverter (MLI), enabling them to be used in the grid-tied mode without the computational burden due to a large number of switching states. The DMPC of LMLI with 289-level output and corresponding 289 control inputs, utilizes a discrete model to predict the future state of the state variable. In order to alleviate the enumeration burden, virtual sectors on a linear scale are introduced, and a general formula is provided to identify the single best state among the 289 states, reducing the time required to find the best optimal state per sampling period. Moreover, the proposed control scheme is independent of objective evaluation.