Model-Free Predictive Current Control of PMSM Based on Coupling-Compensated Ultra-Local Models and Optimized Extended State Observers

Aiming at the strongly coupled nature of surface-unted permanent magnet synchronous (PMSM) and the influence of back electromotive force, the traditional Proportion Integration (PI) control for the current loop has limited bandwidth and insufficient anti-disturbance ability, failing to provide optimal dynamic performance. To address this issue, this paper proposes a model-free predictive control method for PMSM based on an optimized extended state observer (ESO). Firstly, a hyperlocal model considering the cross-coupling effect is established. Secondly, a bandwidth-adaptive ESO is designed to perform real-time observation and dynamic compensation of lumped disturbances, reducing the computational complexity. Finally, experimental validation on a 1.5kW back-to-back power converter platform demonstrates that, compared to conventional PI control, the proposed strategy achieves accelerated dynamic response, substantially suppressed steady-state current oscillations, and consistently low harmonic distortion across varying load profiles. These improvements collectively enhance the operational robustness of motor servo systems, offering a viable solution for high-precision motion control applications.