A Cascaded Band Based Model Predictive Current Control for PMSM Drives

Abstract

Conventional model predictive control for permanent magnet synchronous motor (PMSM) drives often introduces a weighting factor for lowering the switching frequency, which unfortunately leads to the tedious tuning works on weighting factors. Moreover, the computational burden is quite heavy since all the voltage candidates have to be enumerated. In this work, a cascaded band-based model predictive current control is proposed, which can reduce the switching frequency without the use of weighting factor as well as alleviate the computational burden. In this method, the latest applied voltage vector will be adopted to predict the current error prior to the iterative prediction process. Next, a band is defined to compare with the pre-calculated current error to determine whether it is to be maintained. As a result, the iterative cost function minimization turns to be an event triggered process. In addition, the hard switch constraint for the candidate voltage vectors are developed to combine with the cascaded band, which can effectively reduce the iterations and avoid the sub-optimum. The effectiveness of the proposed method is verified on a PMSM platform fed by 3-level T-type inverter.