Resonant frequency, phase and power estimation in a broadband substrate integrated waveguide two- channel T- type power divider using adaptive network-based fuzzy inference system

A novel substrate integrated waveguide (SIW), power divider is proposed. It consists of two channels made by SIW with the same length and width. The bandwidth of 5GHz to 14 GHz is studied here. The propagation constant of the output signals can be adjusted by only four vias in the middle of the output arms. Hence, the position of four metalized inductive vias used here are chosen as a variable ranging from 0 mm to 1 mm and they are shifted in four different positions: up, down, right and left sides. Our studies reveal that moving the vias up and down has no effects on the resonant frequency and other related parameters. However, the values of output powers and phases are changed as the vias are shifted to the right or left sides. Conventionally, artificial neural networks are tested out to obtain resonant frequency, output powers and output phases. But surprisingly, they lead to a promising result. Finally, adaptive network-based fuzzy inference system (ANFIS) is applied in three different steps to obtain resonant frequency, output powers and output phases. As can be seen, ANFIS can determine S parameters as output signals with a high accuracy when the resonant frequency and phases of output ports are given and the average error of less than 10% can be achieved. Nevertheless, estimation of output phases or resonant frequency result in less satisfactory results.