A Fixed Switching Frequency Direct Model Predictive Control for Neutral-Point-Clamped Three-Level Inverters With Induction Machines
Qifan Yang, Πέτρος Καραμανάκος, Eyke Liegmann, Wei Tian, Tobias Geyer, Ralph Kennel, Marcelo Lobo Heldwein
Abstract
This article presents a direct model predictive control (MPC) scheme for drive systems consisting of a three-phase three-level neutral-point-clamped inverter and an induction machine. Even though the discussed MPC algorithm is a direct control strategy, it operates the inverter at a fixed switching frequency, while the output harmonic spectrum of the stator current is discrete, with harmonics at nontriplen, odd integer multiples of the fundamental frequency. As a result, the proposed method achieves similar or superior steady-state behavior than that of modulator-based control schemes. Moreover, thanks to its direct control nature, it exhibits the fast transient responses that characterize direct controllers due to the absence of an explicit modulator. Furthermore, the multiple control objectives of the system, i.e., stator current control and neutral-point potential balancing, are addressed in one computational stage, thus avoiding any additional control loops in a cascaded or parallel structure. This favorable control structure is facilitated by the adopted modeling approach, according to which the system behavior is described by the gradient of the system output. In doing so, not only a simple, versatile system model is derived, but also the direct MPC can be formulated as a constrained quadratic program, which can be easily solved in real time with an in-house solver. The effectiveness of the proposed control scheme is experimentally verified on a 4-kW drive system.