Computationally Efficient and Weighting Factorless Predictive Current Control of Three-Level PMSM Drives by Cost Function Division

The conventional model predictive current control (MPCC) of a three-level neutral-point-clamped (3L-NPC) inverter-fed permanent magnet synchronous (PMSM) drive has a high computational burden because all 27 voltage vectors (VVs) are evaluated in the optimization process. Furthermore, the tuning weighting factor of a cost function that involves several objectives is a challenging operation when employing traditional MPCC. The selection of these weights significantly impacts the control variables, for instance, current, torque, and neutral point voltage (NPV). To get rid of these problems of the conventional MPCC, this article proposes a computationally efficient MPCC method for PMSM drives without a weighting factor. The proposed method evaluates only 6 voltage vectors for current control in two different cost functions, which significantly reduces the computational burden. Additionally, when NPV fluctuates beyond a specified range, another cost function is used for NPV balancing, which evaluates only two small VVs that result in further reduction of the computational load. The complex operation of calculating weighting factors is no longer desirable. As confirmed by simulation analysis, the effectiveness in terms of current, speed, torque, and NPV fluctuation remains unaffected by the removal of the weighting factor and the computational burden reduction, in contrast to the traditional approach.