Adaptive particle swarm optimization based hybrid small-signal modeling of GaN HEMT

Abstract

Based on the GaN HEMT 20-element small-signal model, an improved particle swarm algorithm is proposed in this paper for the optimization of the intrinsic parameters. The conventional Particle Swarm Optimization (PSO) algorithm is prone to fall into local optimum solutions when parameter optimization is performed, in turn to affect the accuracy of experimental results. To solve the above problem, dynamic evolutionary estimation, elite learning strategies and adaptive strategies are added to the traditional PSO algorithm, in turn to yield Adaptive Particle Swarm Optimization (APSO) algorithm in this paper. The experimental results show that the situation of falling into a local optimum for the traditional PSO algorithm is greatly improved by the APSO algorithm, and that the model parameters of the GaN HEMT small-signal model can be extracted more accurately and faster in the frequency range of 0.5–20 GHz using the APSO algorithm.