Skip to Main Content
In this paper, artificial neural network (ANN) approaches to modeling of high-frequency effects of embedded passives in multi-layer printed circuits are presented. Recently developed automatic model generation (AMG) methods for efficient training of ANN models are described, allowing ANN models to automatically learn from electromegnetic (EM) behavior of embedded resistors and capacitors. Through fast and accurate EM-based neural models, we enbable consideration of EM effects in high-frequency and high-speed computer-aided design (CAD), including component's geometrical/physical parameters as optimization variables. Demonstration examples including geometrical/physical-orientated neural models of embedded capacitors and resistors are presented.
Date of Conference: 18-21 April 2007