Real-Time Capable Thermal Model of an Automotive Permanent Magnet Synchronous Machine

Abstract

Excessive temperatures can lead to accelerated aging and irreversible damage in electric machines. Therefore, real-time temperature monitoring is vital for highly utilized electric machines in automotive drives to ensure temperatures are within safe operating limits during operation. Installing temperature sensors on all critical parts would incur too much cost. Hence, model-based real-time temperature monitoring is a preferred solution. Recent publications typically utilize low-dimensional lumped-parameter thermal networks. This paper presents a modeling method for a permanent magnet synchronous machine, where the thermal model is derived using the finite-volume method. The model is calibrated with measurement data. A model-order reduction method is applied, which significantly reduces the computational costs of the model while preserving important (uncertain) parameters such as heat transfer coefficients. Experimental results for different load cycles of the considered machine validate the feasibility and accuracy of the proposed model. Finally, comparing the model to measured temperatures at positions not used for calibration shows that the proposed method accurately captures the temperature distribution in the whole machine without changing the model structure.