Model Predictive Control of Six-Phase Electric Drives Including ARX Disturbance Estimator
Mario Bermúdez, Manuel R. Arahal, Mario J. Durán, Ignacio González‐Prieto
Abstract
Finite-control-set model predictive control (MPC) including virtual/synthetic voltage vectors (VVs) has been recently proposed for the high-performance regulation of multiphase induction motor drives. However, the performance of VV-MPC still deteriorates when the predictive model presents inaccuracies due to simplifying assumptions or erroneous machine parameters. Nonmodeled effects act as disturbances for the control and ultimately reduce the drive performance. From a different perspective, autoregressive with exogenous variable (ARX) models can be used to predict the future state of the drive based on past values of the system without using a physical model. ARX models are included in this article within the VV-MPC scheme to further enhance the predictive capability and control performance by accounting for model mismatches and disturbances. Experimental results confirm that the proposed VV-ARX-MPC can successfully improve the current tracking, reduce the stator copper losses and provide the drive with further robustness against machine parameter variations.