Enhanced Model-Free Predictive Current Control for PMSM Based on Ultra-Local Models: An Efficient Approach for Parameter Mismatch Handling

Traditional model predictive current control (MPCC) is heavily dependent on the accuracy of motor parameters and incurs high computational costs. To address these challenges, this paper proposes an enhanced model-free predictive current control (MFPCC) strategy based on ultra-local models (ULMs). Initially, a Kalman filter (KF) is used to estimate the current gain, while an adaptive sliding mode observer (SMO) is employed to estimate current disturbances. Subsequently, an equivalent transformation of the cost function is carried out in the αβ domain, and the voltage vector combinations are reduced to a single one via sector distribution. Hence, the proposed MFPCC is independent of motor parameters and capable of reducing computational complexity. Simulation and experimental results demonstrate that the proposed MFPCC method significantly improves computational efficiency and the robustness of current prediction, enabling precise current tracking even in the presence of motor parameter mismatches.