Model predictive control of three-level inverter-fed induction motor drives with switching frequency reduction

Abstract

Model predictive control (MPC) has recently emerged as a powerful control method for ac motor drives, featuring quick dynamic response and intuitive concept. However, conventional MPC requires lots of calculation resources due to the enumeration of each possible voltage vector, especially in the case of a multilevel converter. To achieve good performance with low switching frequency for high power drives, the state-of-the-art method uses multi-step prediction, which is even more complicated. This paper proposes a single-vector-based MPC for induction motor (IM) drive supplied by a three-level neutral-point-clamped (3L-NPC) inverter operating at low switching frequency. Instead of torque and flux control in conventional MPC, an equivalent stator flux vector is firstly constructed from the reference value of torque and stator flux, and then the desired voltage vector reference is obtained based on the principle of deadbeat flux control. An optimal cost function is proposed to select the best voltage vector which is closest to the reference voltage vector. Only the voltage vectors producing no more than one jumps in both phase and line voltages are considered as the candidate voltage vectors, which greatly reduces the number of predictions. Furthermore, the suppression of neutral point potential fluctuation and switching frequency limitation are also considered in the cost function. Experimental results are presented to validate the effectiveness of the proposed method at 500 to 760 Hz switching frequency.