Reference Voltage Vector Based Model Predictive Torque Control with RMS Solution for PMSM

Abstract

To reduce the computational burden of a conventional model predictive torque controller (MPTC), a reference voltage vector based MPTC strategy is proposed. The reference voltage vector is obtained from the reference stator flux vector and the reference torque. According to the location of the reference voltage vector, a first optimal vector can be determined in a quite straightforward way, improving the system dynamic performance. Furthermore, in order to decrease the torque and flux ripple, a root mean square (RMS) based solution is employed to generate the reference voltage vector and calculate the duty ratio. This method aims at minimizing the RMS error of flux and torque during the whole control period. Then, the steady state performance is improved. Besides, since the new cost function contains only the reference voltage vector, the weighting factor in conventional MPTC is eliminated. In addition, to keep a balance between the steady state performance and switching frequency, the candidates for the second optimal vector are restricted to a certain scope. Simulations were carried out and the results verified the validation of the proposed MPTC strategy.