Large-signal modeling of HEMT device based on neural network
Lingling Sun, YiXing Liao, Xuefeng Zheng, Zhigang Cheng, Jun Liu, Lei Zhou · 2003
In this article, a neural network model using Levenberg-Marquart algorithm and hyperbolic tangent activation function is developed to simulate three nonlinear parameters for HEMT device. Compared with BP neural network, this ANN model can accelerate convergence. The example shows that the developed ANN model possesses higher accuracy and good generalization. It can be applied to device modeling efficiently.