Ultra-Fast Power Module Inductance Estimation using Convolutional Neural Networks

Abstract

The widespread usage of wide bandgap (WBG) semiconductors forces extra emphasis on the early estimation of the layout parasitic elements. Be it a printed circuit board or a power module, layout optimization is necessary to minimize the negative effects of present inductances. Unfortunately, multiple invocations of inductance extraction software can be timeconsuming. In this work, state-of-the-art convolutional neural networks (CNN) are applied in order to lower the time consumption of inductance estimation without compromising the accuracy.