Low-Complexity Multistep Sequential Model Predictive Current Control for Three-Level Inverter-Fed Linear Induction Machines

Abstract

This paper proposes a low-complexity multistep sequential model predictive current control (M-SMPCC) for three-level inverter-fed linear induction machine (LIM) drives to further improve the steady-state drive performance at low switching frequency. In order to implement multistep MPC for performance improvement, the idea of reducing the search space to fewer voltage vectors (VVs) is adopted to reduce its dramatically increasing complexity. However, there are some problems of the existing simplified methods directly applied for long prediction horizons. First, the reference VV in existing methods cannot consider the performance of long prediction horizons, so a reference VV sequence considering current tracking performance and switching frequency is derived to guide the optimal VV search. Second, with the existing search methods, some VVs that yield better performance in long prediction horizons may be excluded in advance. Hence, a more detailed region division and search method is designed. Moreover, to address the inherent neutral point voltage (NPV) fluctuation in the topology, one sequential optimization strategy is proposed, in which the current tracking performance and NPV balance are guaranteed by two separate cost functions. Finally, the proposed method has validated through comprehensive simulation and experiments, showing lower current harmonics and thrust ripples at low switching frequency.