Model Predictive Torque Control of PMSM Without Weighting Factors Based on MRAS Parameter Identification

The design of weighting factors in the cost function of traditional model predictive torque control (MPTC) is relatively cumbersome, at the same time, the accuracy of the prediction model decreases obviously when the motor parameters are mismatched. Therefore, a model predictive control without weighting factors based on on-line identification of motor parameters is studied. Firstly, the control objectives transformed from torque and flux of traditional MPTC to active torque and reactive torque, since they are of the same dimension the design of weighting factors is unnecessary. Secondly, aiming at the problem of control performance degradation caused by the change of motor parameters in the prediction model, the online identification of motor parameters based on model reference adaptive system is studied, the identification results are applied to the prediction process to avoid the bad influence caused by the parameter variation. The findings from the simulation indicate that the proposed methodology maintains both the dynamic and steady-state performance characteristic of traditional MPTC, while also eliminating the need for weighting factors. Furthermore, it demonstrates an ability to accurately identify motor parameters and enhance control performance in response to parameter variations.