Computationally Efficient Preselection-Based Model Predictive Control for Five-Phase Three-Level Drives

Abstract

Despite the great potential of finite control-set model predictive control (FCS-MPC) techniques, the large calculation load impedes their application on complex drive systems, such as multiphase multilevel drive systems. In this paper, a new pre-selection-based (Pre-S)-FCS-MPC weighting factors-less MPC scheme is proposed for three-level (3L) five-phase (5Φ) drive systems. As an example of 3L drives, the neutral point clamped (NPC) inverter driving a 5Φ PMSM is studied in this paper. The proposed control scheme utilizes all eligible virtual voltage vectors (V3s) as a control set for third-order voltage harmonic elimination. The control set is divided into two groups according to their effect on the DC-link capacitors voltages. The candidate V3s are pre-selected from the proper control-set group to ensure the DC-link voltages balance. In addition, the deadbeat (DB) control operation principles are used to predict the vector components of the reference voltage. Using the voltage reference vector information, only two or three candidate vectors are selected to be examined through a weighting factors-less objective function at every sampling interval. Finally, the dwell times of the applied voltage vectors are optimally redistributed to compromise the control objectives. The proposed Pre-S-FCS MPC results are analyzed and compared to the existing FCS-MPC schemes using MATLAB simulation as well as real experiments.