Multi-Objective Direct Model Predictive Control of Symmetric N-level Multilevel Inverter

This paper presents a modified direct model predictive control of a symmetric N-level multilevel grid-tied inverter. The proposed framework converts the multi-objective optimization problem of reference tracking and switching error control to a single objective in such a way that an analytical formula to calculate the unconstrained solution is obtained. The unconstrained solution can be used to find the optimal switching state using the proposed Generalized Search Algorithm without exhaustive search and objective function evaluation in every sampling interval. Moreover, a wide frequency range weighting factor tuning method is provided based on particle swarm optimization so that the inverter can operate at a desirable average switching frequency. The proposed idea is validated by applying it on two cascaded units of a 17-level switch ladder multilevel inverter, thus giving 289 levels in a grid-tied configuration. The hardware in-loop validation is also provided for the proposed control scheme.