Abstract

Interior permanent magnet synchronous machines (IPMSMs) are characterized by nonlinear torque-current relationships, voltage and current constraints, and parameter variations. However, most optimal torque control methods assume that machine parameters are accurately known. Conversely, existing robust control methods fail to seamlessly integrate all optimal strategies, such as maximum torque per ampere, flux weakening, maximum current, and maximum torque per voltage. This paper proposes an extended state observer-based predictive torque control (ESO-PTC) for IPMSMs that encompasses all optimal strategies while ensuring robustness to variations in inductance and flux-linkage. The conventional dynamic equations and flux-linkage lookup tables are replaced with the ultra-local model and flux-linkage estimator, respectively. The constrained optimal torque control problem is reformulated using the augmented Lagrangian method and solved online with a finite control set approach. The ESO-PTC is computationally efficient, allowing for real-time implementation on a general-purpose off-the-shelf processor. The effectiveness of the proposed method is validated through both experimental results and numerical simulations.