Supervised Machine Learning Assisted-Modeling for Prediction of Geometrical and Electrical Parameters of Printed Antennas

The present era signifies a notable advancement in antenna design and performance optimization through the integration of machine learning techniques. This paper explores the application of supervised machine learning algorithms namely- Random Forest, Ridge and Least Absolute Shrinkage Selector Operator (LASSO) Regression for the parametric prediction of printed antennas. It specifically investigates two categories of antennas: patch antenna without slots and patch antenna with slots. This work emphasizes the prediction of geometrical parameters of both types of antennas using the electrical parameters of the respective antennas. Also, the supervised algorithms are utilized for the prediction of electrical parameters, including single attributes such as resonance frequency of the former and multiple attributes including notched and cut-off frequency points of the latter. The efficacy of the supervised machine learning algorithms is evaluated in terms of mean absolute error (G_MAE) and root mean square error (G_RMSE) metrics.