Thermal Monitoring of Electrical Machines by Linear Differential Equations and Numerical Calibration Procedure

Thermal management becomes increasingly relevant for electric drive systems with high power density. As a core motivation rotating and highly integrated motor components shall be monitored. The enhanced information can be used for thermal protection without additional sensors. In contrast to loss estimation and lumped parameter thermal networks (LPTN) a mathematical approach (MATH) based on smart differential equations is explained in this paper. The algorithms can be implemented into existing control architectures and calculate the thermal behaviour of different hot spots such as permanent magnets and bearings during operation. Beside the concept of this abstract thermal model, the numerical calibration procedure is described. By considering a detailed FEM reference model of the motor a learning drive cycle is defined and simulated. Afterwards the parameters are automatically determined and optimised during calibration. An exemplary IPM synchronous motor and validation drive cycle is taken for evaluating the concept. In comparison to a LPTN as well as the FEM reference model and measurements, the mathematical approach leads to attractive results and effectiveness.