Robust Predictive Current Control Using Luenberger Observer Applied to an Induction Motor for Agricultural Electrical Traction

Induction motors are broadly used in various industrial applications due to their durability, simplicity, and cost-effectiveness. In agricultural electrical traction systems, they are crucial for providing reliable, efficient, and high-torque performance. Their ability to deliver consistent power at varying speeds and loads makes them ideal for such applications, where robust and low-maintenance solutions are essential for operational efficiency. However, parameter mismatches can compromise the performance of the motor control. Advanced control strategies, such as finite control set-model predictive control (FCS-MPC), have been developed to address these challenges. This paper introduces a novel robust predictive current control method for induction motors using an improved Luenberger observer to cope with parameter mismatches. By deriving a discrete-time voltage model from the motor's dynamic model, a Luenberger observer with an inverse linear gain is integrated to predict future stator current and disturbance values, thereby improving the controller's performance under parameter mismatches. Unlike traditional FCS-MPC approaches, this method evaluates the stator voltage in the cost function rather than using current, torque, or magnetic flux. Experimental evaluations with the induction motor under steady-state and dynamic conditions demonstrate the proposed method's superior robustness compared to traditional predictive current control in electrical traction applications. The new method significantly reduces oscillations under substantial variations in inductance and resistance, confirming its effectiveness as a prospective solution.