Model Predictive Overload Control of Induction Motors

In automotive applications, the transient peak torque of electrical machines is essential for the vehicle dynamics. The considerable thermal capacity of the motor allows a much higher transient torque production than the rated torque. However, state-of-the-art overload controllers follow conservative control strategies, which do not consider the actual thermal condition of the machine. The electrical motors are often not fully utilized in this case. This paper presents a model predictive overload control algorithm of an induction motor. The thermal condition of the motor was on-line estimated using a temperature observer based on a lumped parameter thermal network. By means of predicting the thermal behaviour of the motor in advance, the overload controller calculates the current limits dynamically. This overload strategy allows a better utilization of the electrical motor and a guaranteed peak-torque production over the prediction horizon. The algorithms were implemented on a digital signal processor and verified on an induction motor.