Small-Signal Modeling of GaN-on-Diamond HEMT Using ANFIS Method

Abstract

This paper develops and demonstrates an accurate and effective approach for small-signal modeling of Gallium Nitride (GaN) High Electron Mobility Transistor (HEMT) using Adaptive Neuro-Fuzzy Inference System (ANFIS) method. The basic working principle of ANFIS is elucidated. Here, ANFIS is applied on a GaN HEMT device characterized for a broad frequency range of 0.1 GHz - 20 GHz and drain voltage up to 30 V. The developed ANFIS based models have shown extremely good accuracy in terms of common regression metrics such as mean squared error, mean absolute error, coefficient of regression ( <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$R^{2}$</tex> ) etc., where <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$R^{2}$</tex> is achieved around %99.9 for both training and testing datasets. To verify the relevance of this paper with the state-of-the-art, the results of several artificial neural networks based small-signal models of GaN HEMTs under similar operating conditions is compared with the obtained results.