Online Parameter Identification for PMSM Model Predictive Control

A strategy integrating recursive least squares (RLS) parameter extraction is proposed to resolve online acquisition challenges during the model predictive current regulation of permanent magnet synchronous motors. An augmented state forecasting framework is constructed to explicitly account for the transient dynamics of the LC filter. Subsequently, the continuously estimated stator resistance, winding inductance, and rotor flux linkage are embedded directly into the predictive algorithm, ensuring a rigorous integration of active identification and control. Numerical validations demonstrate that the introduced technique facilitates smooth system startups and preserves robust tracking capabilities against sudden load disturbances. Furthermore, the identified variables exhibit reliable convergence, thereby substantiating the theoretical validity and practical feasibility of the developed methodology.