Model Predictive Control of a Switched Reluctance Machine for Guaranteed Overload Torque

Abstract

Electric machines can be overloaded for short periods of time due to their inherently large thermal capacitance. The state-of-the-art overload strategies are often based on static characteristics of a machine. The real-time thermal condition is not considered in such strategies, leaving overload potential partially unused. This paper presents a model predictive overload strategy for a water-cooled automotive switched reluctance machine. The maximum allowed torque is predicted based on a real-time hot-spot temperature estimation, ensuring that the thermal capacitance of the machine is fully exploited. Due to the predictive nature of the algorithm, the peak torque production is inherently guaranteed over the prediction horizon. The proposed strategy is validated on a test bench. The proposed overload algorithm can be applied to other machine types as well.