Using Machine Learning Technology to Online Predict the Maximum Common Mode Current of Three-phase Motor Drive Inverter

To reduce the common-mode voltage (CMV) in the PWM-based motor drive system, many CMV reduction methods have been proposed. However, the performance of such methods has limitations such as only being implemented on particular operating conditions with fixed switching frequency or PWM patterns and relying on the simulation or experimental data. This paper explores machine-learning-based methods to actively evaluate the CM performance. Machine learning methods are employed to actively analyze three popular PWMs (SVPWM, AZSPWM, and DPWMMin) on-chip. In this way, we can online determine the best PWM pattern and switching frequency with a minimum requirement of computation resources based on the torque and speed command.