Online temperature estimation of an automotive switched reluctance motor using space-resolved lumped parameter network

The critical components in electric machines must be protected against overheating. Overheating shortens the life span of the machine and causes machine failure in extreme cases. This paper presents an online temperature estimator for an automotive switched reluctance machine based on a space-resolved lumped parameter thermal network. The losses are calculated online using a pre-measured loss map and a set of loss-splitting functions. The effects of the non-ideal thermal sensors are taken into account. The parameters of the temperature estimator are determined by means of a global parameter identification procedure. The results given by this model show a temperature estimation error less than 7 °C over the full speed and torque range. The ability to accurately predict hot-spot temperature provides evidence for the use of advanced thermal protection and allows for an optimized utilization of switched reluctance machines.