Small signal behavioral modeling technique of GaN high electron mobility transistor using artificial neural network: An accurate, fast, and reliable approach

摘要

This article reports a comparative study of two artificial neural network structuresand associated variants used to describe and predict the behavior of 2 × 200 μm2GaN high electron mobility transistors (HEMTs), utilizing radiofrequency characterization.Two architectures namely multilayer perceptron and cascadefeedforward, have been investigated in this work to develop the behavioral model.A study is conducted utilizing the two architectures, all trained using Levenberg-Marquardt, in terms of accuracy, convergence rate, and generalization capabilityto develop the behavioral model of GaN HEMT. However, to ensure the robustnessof the model, accuracy, convergence rate, time elapsed, and generalizationcapability of the proposed model is also tested under couple of training algorithms,activation functions, number of hidden layers and neuron embeddedinside it, methods for initialization of weights and bias and certain other vitalparameters playing vital role in influencing the model accuracy and effectiveness.An excellent agreement found between measured S-parameters and the proposedmodel proves the effectiveness of the proposed approach and excellent predictionability for a sweeping multibias set and broad frequency range of 1 to 18GHz.Moreover, a very good generalization capability is also recorded under variationof crucial parameters of GaN HEMT-based neural model.