Small-signal behavioral-level modeling of InP HBT based on SO-BP neural network

Abstract

In this paper, Snake Optimizer-Back Propagation (SO-BP) neural network algorithm has been proposed to model the small-signal characteristic of InP heterojunction bipolar transistors (HBTs). The SO algorithm can effectively improve the problem of local optima encountered by the BP neural network algorithm. The InP HBT small-signal models have been constructed by using Back Propagation (BP) algorithm, the Genetic Algorithm-Back Propagation (GA-BP) algorithm and the Snake Optimizer-Back Propagation (SO-BP) algorithm. The experimental results verify that the proposed SO-BP neural network algorithm model has the lowest error rate between simulated and measured data. Hence, the SO-BP algorithm is a more effective approach for modeling the small-signal characteristic of InP HBT. The accurate small-signal behavioral-level model of HBT has significant reference value for the application of microwave and millimeter-wave integrated circuits (MMICs).