Estimação do comprimento de antenas de microfita utilizando Support Vector Regression

Abstract

In this paper, a Machine Learning technique called Support Vector Regression is applied for regression of squared microstrip patch antennas data for operation under its dominant mode 𝑇𝑀10. The simulation data was gathered through Ansys Electronics Desktop 2022, using the high frequency simulation module HFSS. Simulation results are given in both S and Z parameters for 2 different situations. Firstly, the antenna length 𝐿 is changed from 10 mm to 40 mm in steps of 1 mm. After it, the antenna behaviour is simulated and the frequency range is fixed from 1 GHz to 10 GHz, getting 𝐿 values that satisfies the specified frequency range. Both steps are simulated for 5 different dielectric permittivity 𝜖. With these informations, a database is created, and it will contain data that will be cleaned, sorted out and labeled, then feeded to a Support Vector Regression (SVR) algorithm for training and validation. The generated model is capable of predicting, accurately, the value of 𝐿 for any 𝑓𝑟 within the frequency range established.