Comparative Analysis of Fixed-Parameter and UKF-Based Adaptive M²PC-Controlled Induction Machines Under Parameter Variations

This paper presents a simulation-based comparative analysis of Modulated Model Predictive Control ($M^{2} P C$) strategies under parameter variations in induction motors. Two control schemes are evaluated: one using fixed (nominal) parameters and another integrating an Unscented Kalman Filter (UKF) for online estimation of magnetizing inductance ($\mathbf{L m}$) and rotor resistance ($\mathbf{R}_{\mathbf{r}}$). Results have shown that parameter mismatch significantly affects electromagnetic torque production, flux regulation, and power quality, leading to performance degradation. The UKF-based controller effectively compensates for these deviations, maintaining accurate torque and flux control. These findings are particularly relevant for electric vehicle applications, where direct torque control is essential. In such systems, unaccounted parameter variations can lead to torque deficits and loss of drive accuracy. The study demonstrates that integrating UKF into model-based control enhances system robustness, efficiency, and torque delivery under dynamic conditions.