This paper proposed a model-free predictive control based on a maximum torque per ampere. The dynamic model of the permanent magnet synchronous motor is used to derive the expression of electromagnetic torque. Using the maximum torque per ampere control method the minimum current reference is derived. To achieve a robust performance under parameter changes and external disturbances influencing the permanent magnet synchronous motor, a model-free predictive control based on an ultra-local model and an extended state observer is proposed. The simulation results showed that compared with the classic linear controller, the proposed control strategy ensures smaller speed and torque ripples and better robustness under parameter mismatches and disturbance.

Model-free predictive current control, extended state observer, maximum torque per ampere, ultra-local model