Robust Online Parameter Estimation of M2Controlled IM Using Unscented Kalman Filter

This study presents an the Unscented Kalman Filter (UKF) based online parameter estimation strategy for an induction machine (IM) controlled by a Modulated Model Predictive Control ($M^{2} P C$) approach. Given the nonlinear nature of IMs and the UKF’s enhanced capability in handling nonlinear systems, the proposed method focuses on estimating magnetizing inductance and rotor resistance alongside stator currents and rotor fluxes in the dq-reference frame. Unlike prior studies, which mainly utilize the UKF for state estimation and employ conventional control methods such as Field-Oriented Control (FOC) or Direct Torque Control (DTC), this work integrates parameter estimation into an $\mathbf{M}^{\mathbf{2}} \mathbf{P C}$ framework to improve control robustness under parameter variations. The proposed estimation-control structure is evaluated in both constant torque and constant power regions. Simulation results confirm that the UKF accurately estimates the IM parameters with less than 10% error, enabling robust and highperformance control across a wide operating range.