Direct Voltage-Selection Based Model Predictive Direct Speed Control for PMSM Drives Without Weighting Factor

To reduce the computation burden and eliminate the weighting factor in conventional speed prediction control, a direct voltage-vector selection-based model-predictive direct-speed control (MP-DSC) method is proposed in this paper. An extended sliding-mode load-torque observer is designed to observe the motor speed and the load torque, which improves the robustness of the system and reduces the influence of measurement noise. Then, the reference voltage vector that includes the system speed and current information is predicted based on the deadbeat control principle. This reference voltage vector is used to construct a cost function, which only includes the error between the reference voltage vector and the candidate voltage vector. Thus, the weighting factor in conventional MP-DSC method is avoided in the proposed method. In addition, a voltage-vector selection method is developed to quickly determine candidate voltage vectors and to ensure the current does not exceed current limit. Finally, the proposed MP-DSC method is experimentally compared with a model-predictive torque and speed control.