Modeling of GaN HEMTs Based on BP Neural Networks

Gallium nitride (GaN) high electron-mobility tran-sistors (HEMTs) serve as an important link in Radio Frequency (RF) circuits, so its accurate modeling becomes crucial. Tra-ditional modeling is based on physical formula, which is not only inconvenient to calculate but also needs to consider a large number of influencing factors. With the development of Artificial Intelligence (AI), this paper proposes a new modeling method for GaN HEMTs based on Back Propagation (BP) neural networks. The method eliminates the complex physical computation steps and trains the neural network only by using actual input and output data to obtain the model. Compared with the traditional modeling, the method eliminates the com-plex physical formula calculation step, which is not only simple but also more accurate. The experimental results show that the model has excellent performance on both large-signal and small-signal models of GaN HEMTs.