Multi-surface Permanent Magnet Synchronous Motor Temperature Estimation based on Automate Machine Learning Approach

The Permanent Magnet Synchronous Motor (PMSM) is a high-performance motor that is primarily utilized to power the motor of an electric vehicle, a new generation of transportation. Since temperature affects PMSM, monitoring real-time temperature using direct temperature detection via sensors complicates and increases the cost. As a consequence, the objective of this research aimed to use artificial intelligence with an automated machine learning technique to estimate three PMSM component temperature values, including stator winding, stator teeth, and the permanent magnet, using the same nine input features: ambient temperature, coolant temperature, direct-axis voltage, quadrature-axis voltage, motor speed, torque, direct-axis current, quadrature-axis current, and stator yoke temperature. As a result of this research, stator winding and stator tooth have prediction values and waveforms that are close to validation values, with Mean Absolute Errors of 4.651 and 1.902, respectively. On the other hand, permanent magnet prediction has a remarkably different value and waveform with validation on 8.054 Mean Absolute Error.