MODEL-FREE PREDICTIVE CURRENT CONTROLLER WITH ULTRA-LOCAL MODEL UTILIZING A NOVEL OBSERVER FOR DUAL THREE-PHASE PERMANENT MAGNET SYNCHRONOUS MACHINE

Abstract

Permanent Magnet Synchronous Machines (PMSMs) are widely adopted in high performance applications such as electric vehicles due to their efficiency and power density. However, advanced current control methods like Deadbeat Predictive Current Control (DPCC) and Model Predictive Current Control (MPCC) depend heavily on accurate machine parameters, making them vulnerable to parameter mismatches and external disturbances. To address this limitation, this thesis proposes a novel Model Free Predictive Current Control strategy integrated with a Nonlinear Disturbance Observer (MFPCC-NDOB) for Dual Three-Phase PMSM (DTP-PMSM) systems. The proposed method employs an ultra-local model that eliminates the need for precise machine parameters by continuously estimating unknown dynamics in real time. The nonlinear disturbance observer enhances estimation accuracy, and its integration into a deadbeat control structure ensures fast dynamic response and high tracking precision. Compared with traditional DPCC and conventional MFPCC approaches, the proposed method significantly improves steady-state accuracy, dynamic stability, and robustness to parameter variation. Simulation and experimental results on a 9 kW DTP-PMSM validate the controller’s superior performance under both nominal and mismatched conditions.