DC-IV model modeling method of GaN HEMT transistor

Abstract

The invention discloses a DC-IV model modeling method for a GaN HEMT transistor, and the method comprises the steps: obtaining the original DC-IV characteristic data of the GaN HEMT transistor through testing the DC-IV characteristic of the GaN HEMT transistor, carrying out the normalization processing, constructing a full-connection feedforward neural network model, and carrying out the modeling of a full-connection feedforward neural network model. Taking data obtained by performing normalization processing on original DC-IV characteristic data of the GaN HEMT transistor as training data to train the full-connection feedforward neural network model, so that a DC-IV model of the GaN HEMT transistor can be obtained; the method has the advantages that the full-connection neural network model serves as a carrier, internal structure information of the GaN HEMT transistor is not needed, modeling can be achieved only through DC-IV test data of the GaN HEMT transistor, the convergence speed is high, the generalization ability is high, the DC-IV characteristics of the GaN HEMT transistor can be accurately simulated and predicted, the modeling period is greatly shortened, and meanwhile the modeling precision is improved.

Description

technical field [0001] The invention relates to a DC-IV model modeling method of a GaN HEMT transistor, in particular to a neural network-based DC-IV model modeling method of a GaN HEMT transistor. Background technique [0002] Due to its unique characteristics, such as high breakdown voltage, high electron mobility, high gain, low switching speed, wide energy bandgap and high operating frequency, GaN HEMT transistors have been widely used in 5G communications, new radars and power electronics. It is widely used and has high research and engineering application value in the field of microwave power devices including 5G power amplifiers. In order to maximize the excellent characteristics of GaN HEMT transistors, it is particularly important to develop an accurate device model. The more accurate the device model of GaN HEMT transistors, the higher the reliability of the circuit design based on the transistors. [0003] The output characteristics of transistors (also known as ...

AI Technical Summary

Problems solved by technology

[0004] Most of the existing transistor models are based on improved models such as the Angelov / Chalmers model, the Curtice model, and the Materka model. These models include factors such as the trap effect and harmonic characteristics of the transistor, and are not accurate enough to characterize the DC-IV characteristics of the transistor, and the model exists Problems such as poor generalization ability and slow convergence speed

Images