Temperature Estimation of Permanent Magnet Synchronous Motors Using Support Vector Regression

In recent years, permanent magnet synchronous motors (PMSMs) are widely used in electric vehicles due to the high-power density. The increasing power density of PMSMs can result in the rapid temperature rise, which leads to a negative impact on the machine. The high temperature may lead to weakened winding insulation in the stator slots and the irreversible demagnetization of permanent magnet. In order to avoid overheating running, massive methods aiming to estimate the temperature inside PMSMs are investigated. However, traditional methods like Lumped Parameter Thermal Networks (LPTNs) require massive domain knowledge and accurate parameters. Therefore, a model-free data driven approach based on support vector regression is proposed to estimate the temperature inside the PMSM in this paper.