Modeling of SiC MESFETs by Using Support Vector Machine Regression

Abstract

In this paper, a support vector machine (SVM) regression approach is introduced for modeling of field effect transistors (FETs). Benefits to the good generalization ability of SVM, a SVM regression (SVR) model is established using a set of training and testing data, which is produced by simulation using an available empirical model of SiC MESFETs. Experimental results show the SVR model has good ability in predicting electrical performance of FETs.