부산대학교 도서관

In this work, a novel multiobjective voltage-vector-based (VVB) finite control set (FCS) model predictive control (MPC) for a permanent magnet synchronous machine drive fed by a three-phase five-level (3P-5L) cascaded H-bridge multilevel inverter (MI) is proposed. This algorithm aims to overcome the main issues relative to MPC implementation detected in the scientific literature for electric drives fed by cascaded H-bridge MIs (CHBMI). In detail, the goals are the minimization of computational cost by reducing the number of required predictions, the minimization of the switching devices state transitions, i.e., the switching losses minimization, and the common mode voltage (CMV) reduction. These goals are fulfilled through an offline optimization process; thus, no additional terms and weighting factors to be tuned are required for the cost function. Experimental validations are presented to prove the effectiveness of the proposed approach. In detail, an accurate electric drive performance comparison, both in steady state and dynamic working conditions, is carried out when the proposed VVB MPC and the cell-by-cell-based MPC are adopted. As comparison tools, current and voltage total harmonic distortion, apparent switching frequency, CMV amplitude, and torque ripple are adopted.