Abstract
Thermal effect represents one of the main challenges for GaN power transistors. In addition to ambient temperature, the power dissipation-induced self-heating represents the main source of heating. This electrothermal effect must be accounted for by the transistor model to provide reliable circuit design. This paper makes use of a single-hidden layer Artificial Neural Network (ANN) to model the drain current. A recently developed global optimization approach is used to train the model and optimize its weights and biases. The model is applied to $8 \times 125-\mu \text{m}$ and $16 \times 250-\mu \text{m}$ GaN High Electron Mobility Transistors (HEMTs) and compared with measurements. Excellent results obtained validate the effectiveness of the ANN-based modeling and optimization methods of power transistors. The main advantage of the approach is its technology independence; thus, it could be applied to a wide range of microwave power transistors.