Selective Voltage Vector Based Model Predictive Control for Three- Level Active- NPC Inverter

Abstract

The stability issue is a major hurdle for the practical model predictive control (MPC) scheme due to its non-linear nature and parameters dependency. Moreover, the computational burden limits the use of MPC in the multilevel inverter as is it has a large number of voltage vectors and requires high-speed processors. The conventional MPC algorithm evaluated all the voltage vectors of multilevel inverters during the prediction and actuation process which creates severe computational complexity. This paper proposes a selective voltage vector-based model predictive control (SV-MPC) scheme for three-level active-neutral-point-clamped (3L-ANPC) inverter. The control law of the proposed SV-MPC scheme is derived by satisfying the Lyapunov stability criteria which use inverter's discrete behavior and directly applies the finite-state switching actions for the efficient operation of the 3L-ANPC inverter. In case of the 3L-ANPC inverter, the conventional MPC requires to evaluate all 19 voltage vectors (116 switching states) which is reduced to only 6 voltage vectors (44 switching states) by the proposed SV - MPC scheme. This reduction in evaluating voltage vectors improves the execution efficiency by 39 % compared with the conventional voltage-mode MPC. The performance of this proposed SV-MPC for 3L-ANPC inverter is investigated with a MATLAB/Simulink model and the results are quite satisfactory. On the other hand, another significant contribution of this paper is to design a stable and nonlinear control law for the 3L-ANPC inverter. The stability of the proposed SV-MPC scheme is verified with phase-plane analysis and parameter variations of the 3L-ANPC inverter system.