Model Predictive Torque Control of Synchronous Machines Without a Current or Stator Flux Reference Generator

Conventional model predictive torque control (MPTC) of synchronous machines (SMs) relies on a current or stator flux reference generator to attract the d-axis current or stator flux to desired operating points while minimizing the torque error. However, using a reference generator has two issues: 1) an additional demanding optimization is required to obtain the reference generator, and 2) the additional optimization restricts the degrees of freedom (DOF) for MPTC to determine the operating points of SMs. Therefore, this study presents an MPTC scheme that does not rely on a current or stator flux reference generator. To this end, first, a novel MPTC problem is formulated to minimize a performance index while satisfying the torque, voltage, and current constraints. The performance index, e.g., copper loss or inverter loss, determines how the MPTC utilizes the DOF in the torque control. Then, the proposed MPTC problem, a nonlinear optimization, is solved based on the finite control set (FCS) using the Augmented Lagrangian method. Simulation results obtained using a 50-kW SM show that 1) the proposed MPTC guarantees optimal operation under all operating regions without a reference generator, and 2) it can significantly enhance efficiency by utilizing its DOF in determining the operating point.