Ultra-Local Model Based Sensorless Predictive Control for Induction Motor Drives

Abstract

Sensorless model predictive control (MPC) of induction motors is highly sensitive to parameter uncertainties and speed estimation errors, which may degrade system performance and stability. To overcome these limitations, an ultra-local model based sensorless predictive control strategy is developed in this study. The induction motor is first modeled in the stationary two-axis reference frame, and the implementation process of the predictive control algorithm is described. To eliminate the need for a mechanical speed sensor, a full-order adaptive observer is designed to reconstruct rotor speed and flux linkage information in real time. Considering the influence of parameter variations, external disturbances, and modeling inaccuracies during operation, an ultra-local dynamic representation is introduced to capture the system behavior without relying heavily on precise motor parameters. Furthermore, a sliding mode observer is employed to estimate and compensate for lumped disturbances, thereby improving the disturbance rejection capability and robustness of the control system. The proposed approach is validated through Matlab/Simulink simulations under different operating scenarios, including parameter perturbations, speed changes, and load disturbances. The simulation results indicate that the proposed method can accurately reconstruct rotor speed while preserving desirable dynamic and steady-state characteristics. Compared with conventional sensorless MPC schemes, the proposed control strategy demonstrates enhanced tolerance to parameter mismatches and stronger resistance to external disturbances. These findings confirm the effectiveness of the proposed approach and its potential application in high-performance sensorless induction motor drive systems.