Smart Voltage Vectors for Model Predictive Control of Six-Phase Electric Drives
Ángel González-Prieto, Ignacio González‐Prieto, Mario J. Durán
Abstract
Model predictive control (MPC) has been recently in the limelight within the field of multiphase machines and drives because it can naturally exploit some of their inherent advantages. Unfortunately, the phase current quality provided by standard MPC is unacceptable when the values of the stator resistance and leakage inductance are low. Aiming to overcome this disadvantage, different MPC strategies using synthetic/virtual voltage vectors (VVs) have been developed. Nevertheless, static VVs are created offline, and consequently, they cannot be adapted to the operating point. This limitation has been recently mitigated with the creation of dynamic virtual voltage vectors (DVVs), where VVs are generated online combining two different switching states per sampling time. Despite the current quality improvement using DVVs, the refinement of the voltage output is still limited, and consequently, the solution is suboptimal. This article suggests the use of smart voltage vectors (SVVs) that are obtained by composing the three best voltage vectors in a smart proportion during the sampling period. The calculation of the duties is performed with an online smart search that allows the real-time implementation of this MPC version. The capability of the proposed SVV-MPC to enhance the current quality indices is validated with experimental results.