Machine learning based Pulsed IV behavioral model for GaN HEMTs

Abstract

A novel Pulsed IV (PIV) behavioral model for Gallium Nitride (GaN) High Electron-Mobility Transistor (HEMT) based on machine learning technique is described in this paper. As one of the core methods of machine learning technique, Bayesian inference is used to build this new model. The performance of the proposed model is validated through experimental test examples. The model can very accurately predict the PIV curves of a 10W GaN HEMT at different pulse widths and duty cycles from isothermal up to the safe-operating area limit, with high voltage drain-source pulses, indicating that the proposed model can include both the thermal and dispersion effect in it precisely and effectly.