Nonlinear Semiconductor Device Modeling using Neural Networks

Abstract

This paper describes the nonlinear semiconductor (transistor) of small and large signal modeling using a single neural network. Multilayer perceptron (MLP) with back-propagation (BP) learning is adopted in this work to model the drain current (I_D) and the transconductance (g_m) of the transistor. MLP modeling performance in terms of mean square error (MSE) and complexity of the network are illustrated briefly. Artificial neural network (ANN) model outcome for nonlinear function estimation (I_D) as well as its derivative (g_m) shows good agreement with the expected behavior.

Keywords: Small and large signal modeling, ANNs, MLP