Robust Predictive Current Control of IMs With MRAS-Based Parameter Estimation

Induction motors are extensively employed in numerous industrial and commercial applications due to their robustness, simplicity, and cost-effectiveness. However, the motor performance is often influenced by parameter mismatches, which can degrade control precision. Advanced control techniques like finite control set-model predictive control (FCS-MPC) address these challenges by utilizing system models to predict future behaviors and select optimal control actions. To further enhance the robustness of predictive control, this paper introduces a novel model predictive current control framework for induction motors in the stationary reference frame. It incorporates a Model Reference Adaptive System (MRAS) observer for real-time parameter estimation, targeting mismatches in inductance and resistance. Unlike conventional FCS-MPC methods, this approach employs stator voltage rather than current, torque, or flux as the optimization metric in its cost function. By leveraging the MRAS observer, the proposed method dynamically aligns model parameters with real-time motor conditions, mitigating the effects of parameter variations and external disturbances. Experimental evaluations on an induction motor under dynamic and steady-state conditions confirm the superior robustness and tracking performance of this approach compared to traditional predictive control strategies. The integration of the MRAS observer demonstrates significant improvements in maintaining control accuracy even under substantial parameter fluctuations, underscoring its potential for industrial application.