Abstract

Ultralocal is always used to solve the problem of weak robustness in predictive control to resist the influences caused by some time-varying physical parameters in the motor driving system. However, the achieved model cannot reflect the motion characteristics of the motor accurately and timely. Considering the influences on the model accuracy caused by nonlinear terms, a model-free predictive current control (MF-PCC) using extended affine ultralocal is proposed in this article for solving these issues. A data-driven model of extended affine ultralocal is built containing a two-order term based on affine arithmetic, and all coefficients in the model are online estimated by the recursive least square algorithm, including the input gain in the MF-PCC strategy based on the conventional ultralocal. The proposed method is completely independent from prior knowledge of the physical parameters of the controlled system. The proposed method is applied to a permanent magnet synchronous motor driving system, and the simulation and experimental results demonstrate the effectiveness and advantages including improved current quality and robustness compared with the MF-PCC strategy based on the conventional ultralocal model.