High-Frequency Modeling of Star-Connected Induction Motors Using Multilayer Perceptron

This letter presents an easy-to-use accurate high-frequency (HF) behavioral model of star-connected induction motors with a simple and straightforward parameterization process for conducted electromagnetic interference (EMI) analysis. The HF model contains nine circuit elements per phase with a simple structure. These circuit elements are determined using a multilayer perceptron (MLP) neural network, where only the individual phase-to-frame impedance is needed for training purposes. Compared with parameterization methods in conventional motor modeling, the MLP significantly simplifies the parameterization process. Using two different induction motors (1.5 hp and 7.5 hp) as case studies, experiments verify the capability and accuracy of the HF model to predict the motor common-mode (CM) and differential-mode (DM) impedances in the HF range of 150 kHz to 30 MHz.