Permanent magnet synchronous motor (PMSM) has become a core component of complex electromechanical systems such as electric vehicles, new energy urban rail vehicles, and wind power generation due to its advantages of high efficiency, high power density, and high torque density. The model predictive control strategy based on the mathematical model of the controlled object has been widely applied in PMSMs. However, there is inevitably a mismatch between the actual parameters of the motor system and the application parameters of the model predictive controller, which seriously affects the performance of the predictive control system. This paper proposes a novel model-free stator flux sliding mode control (MF-FSMC) method to achieve high-performance control under parameter mismatch.

The model-free flux sliding mode controller is adopted to replace the model predictive control algorithm that relies on the controlled object. First, a mathematical model of PMSM is established under parameter mismatch, and an ultralocal stator flux linkage model considering parameter mismatch is constructed under a rotating coordinate system. The novel MF-FSMC method is proposed, a model-free flux sliding mode controller based on a novel reaching law is designed, and a one-beat speed predictive controller is constructed. Finally, the composite integral sliding mode disturbance observer online estimation strategy is proposed, which can effectively observe the unknown disturbance part in the ultralocal model of PMSM under parameter mismatch.

Simulation and experimental results show that the proposed method can effectively improve the steady-state performance, significantly reduce the stator flux and torque ripple, and enhance the robustness and anti- interference performance of the PMSM system under parameter mismatch. The designed composite integral sliding mode disturbance observer can accurately observe the stator flux values and unknown disturbances on the dq axis. Compared with conventional model predictive control methods, the proposed MF-FSMC method can reduce torque ripple from 75 N·m to 45 N·m in the case of flux linkage parameter mismatch. In addition, with the proposed MF-FSMC method, the distortion of stator current has also been significantly improved, and the static error of stator flux linkage has been reduced from 0.35 Wb to ±0.01 Wb. In the case of inductance parameter mismatch, the proposed MF-FSMC method can reduce torque ripple from 110 N·m to 80 N·m and the fluctuation value of stator flux linkage error from ±0.235 Wb to ±0.02 Wb.

The MF-FSMC method proposed can obtain the following conclusions: (1) The composite integral sliding mode disturbance observer can accurately observe unknown disturbances and stator flux linkage, effectively enhancing the robustness of predictive control systems under parameter mismatch. (2) The proposed model-free stator flux sliding mode control method designs a model-free flux sliding mode controller in the inner loop of the controller and constructs a one-beat speed predictive controller in the outer loop of the controller. The proposed MF-FSMC method can effectively improve the control accuracy of stator flux, significantly reduce the stator flux/torque ripple of the motor, and ensure the strong robustness of the predictive control system under parameter mismatch.