Improved Deadbeat Predictive Current Control With Embedded Resonant Polynomial and Disturbance Observer for PMSM Current Distortion Rejection

Deadbeat predictive current control (DPCC) holds the superiorities of excellent current tracking ability and high control bandwidth. However, the DPCC performance is negatively impacted by parameter mismatch and inverter nonlinearity, resulting in current distortion. To address these issues, a reduced-order generalized proportional integral (GPI) observer-based robust resonant DPCC (RRDPCC) method is presented in this article for permanent magnet synchronous motor (PMSM). First, the non-periodic uncertainties generated by parameter mismatch and the periodic current harmonics induced by inverter nonlinearity are analyzed. Then, an accurate PMSM model is given considering the periodic and non-periodic disturbances. Second, a resonant polynomial is embedded into the current prediction model, which is proven to effectively reject periodic current harmonic. Third, a reduced-order GPI observer is utilized to track the lumped disturbance induced by parameter mismatch. Additionally, the stability analysis of the observer is also provided. Consequently, the proposed RRDPCC is robust to both non-periodic and periodic distortions. Ultimately, the experimental results confirm the feasibility of the proposed distortion rejection strategy.