Model Predictive Pulse Pattern Control for the Five-Level Active Neutral-Point-Clamped Inverter

Abstract

In this paper, the recently introduced model predictive pulse pattern control (MP3C) strategy is adapted to the ACS 2000 medium-voltage (MV) drive of ABB. The drive system consists of a five-level active neutral-point-clamped (ANPC-5L) rectifier, an inverter, and an induction machine (IM). The inverter is fed with offline-computed optimized pulse patterns (OPPs) that produce minimum harmonic distortion in the stator windings of the ac machine. An optimal stator flux trajectory is calculated from these OPPs, and a trajectory controller tracks it in real time. In the proposed approach, trajectory tracking is based on model predictive control: a constrained optimal control problem is formulated and solved in real time in a computationally efficient manner. An event-based prediction horizon is employed in order to ensure fast tracking of the stator flux trajectory. The advantages of the proposed method are optimal steady-state behavior in terms of harmonic distortion and fast torque response. The method was tested on an MV ANPC-5L inverter coupled to a general-purpose 1.21-MW IM. Experimental results were obtained from this industrial setup, and they are presented in this paper to demonstrate the high performance of MP3C.