Model-free predictive current control for a SynRM drive based on an effective update of measured current responses

Abstract

Predictive current control schemes strongly rely on the knowledge of the plant model. The accuracy of the current prediction could be affected by parameters variation or mismatch, non idealities and other model inadequacies. In Synchronous Reluctance Machine this effect could be particularly critical since its inherent intense iron saturation causes the variation of the indunctances in a wide range. For this reason, the nominal model could be unsuitable for current prediction and could lead to a severe deterioration of the drive performance. In this paper a novel Model-Free Predictive Current Control is presented. The current variations related to the feeding voltage vectors are recorded on-line and stored in a Lookup Table. The current prediction is later obtained by accessing the LUT in the position related to the considered voltage vector. The novelty of this work is the updating method for the Lookup Table. The proposed technique allows reconstructing all the current variations using only the responses related to the three most recent voltage vectors. The result is an improved Model-Free predictive current control in which the reliability of the prediction is guaranteed by the frequent LUT update while always observing the optimal control given by the cost function minimization.