This paper introduces an estimator for the online identification of 𝐿 𝑞 that obviates both the integration stage and dependence on stator resistance. The proposed method is fundamentally a flux-based Model Reference Adaptive System (MRAS) but utilizes the natural excitation provided by a standard Space-Vector Pulse Width Modulation (SV-PWM) pattern. By oversampling the stator current within a single PWM switching period and leveraging the averaging properties of the applied voltage vectors, the estimator directly reconstructs the q-axis flux linkage without requiring explicit integration of terminal voltages. Consequently, the derived adaptation mechanism is rendered independent of stator resistance and immune to DC offset integration errors. A comprehensive comparative simulation study, implemented in MATLAB/Simulink, validates the estimator's performance against the classical fluxintegration-based MRAS across a rigorous set of operational scenarios. Results unequivocally demonstrate the superior accuracy, dynamic response, and parametric robustness of the proposed PWM-based method, particularly under stator resistance variation, speed sensor inaccuracy, and operating point transitions, including zero-speed crossing and regeneration modes.

This paper presents a predictive model reference adaptive system (MRAS) speed estimator for Permanent Magnet Synchronous Machines. The speed estimator is based on the finite control set model predictive control (FCS-MPC) principle. A search method is utilized to find the best estimated speed at each sampling interval that produces the smallest speed estimation error signal. This method replaces the PI controller which is typically employed by MRAS estimators. The speed estimator has been experimentally tested using a 2.1kW PMSM. Results show improved performance, especially during fast load transients for the novel method compared to the PI-based MRAS method.