Affine Ultra-Local Based Model-Free Predictive Control on PMSM Drives with $\varepsilon$-Approximation State Compensation

The accuracy of a data-driven model directly affects the control performance of model-free predictive control (MFPC). To take into account the affine motion of the plant, a model-free predictive current control (MF-PCC) strategy is proposed in this paper and applied to a permanent magnet synchronous motor (PMSM) driving system. To enhance model accuracy, we adopt an affine ultra-local model, which is designed to accurately represent the system. Online estimation of all affine operators is performed using the recursive least squares (RLS) algorithm to ensure a robust fit. To improve control performance, we integrate a state compensation technique with the model. The compensation gain is selected using the $\varepsilon$-approximation approach to meet terminal constraints. The stability of the proposed method is theoretically analyzed, and its effectiveness is verified through experimental results. The proposed method offers improved model accuracy and current quality, along with enhanced robustness compared to conventional methods.