Permanent Magnet Synchronous Machine Temperature Estimation using Low-Order Lumped-Parameter Thermal Network with Extended Iron Loss Model

Since temperature rise in electric machines is mainly
due to power losses during electro-mechanical power conversion, temperature estimation is highly attached to power loss modelling. In this contribution, an extended iron loss model is introduced with a direct identification methodology in the context of temperature estimation. The iron loss model is implemented as part of a fourth-order lumped-parameter thermal network (LPTN), which is parametrised using empirical measurements and global identification. Once parameters are identified using training data, the LPTN model is validated using three unseen profiles cross-validation. Satisfactory estimation is achieved with the average mean squared error of 2.1 K2 and the error bias close to zero.