Open AccessDesign of Microstrip Filter by Modeling with Reduced Data
Author(s) -
Ahmet Uluslu
Publication year - 2021
Publication title -
applied computational electromagnetics society journal
Language(s) - English
Resource type - Journals
eISSN - 1943-5711
pISSN - 1054-4887
DOI - 10.13052/2021.aces.j.361109
Subject(s) - microstrip , cutoff frequency , electronic engineering , microwave , filter (signal processing) , computer science , frequency band , low pass filter , frequency response , engineering , telecommunications , bandwidth (computing) , electrical engineering , computer vision
Many design optimization problems have high-scale problems that require the use of a fast, efficient, accurate, and reliable model. Recently, artificial-intelligence-based models have been used in the field of microwave engineering to model complex microwave stages. Here, an eight-layer symmetrical microstrip low-pass filter (LPF) is modeled using a multi-layer perceptron (MLP) with reduced data with Latin hypercube sampling. It is used to obtain target−-test relationships in the MLP model along the frequency band whose electrical length in each layer determines the performance of the microstrip filter. Electrical length lower and upper limits were preferred in the widest range. The study presents the design and analysis of a non-uniform symmetrical microstrip LPF with a cutoff frequency of 2.4 GHz. Next, different network models are compared to find the variation of the non-uniform microstrip LPF around 2.4 GHz along the specified frequency band S1111 and S2222 (dB) for different electrical lengths. It has been observed that the network models of the microstrip LPF are both more computationally efficient and as accurate and reliable as the electromagnetic simulator.