Design and Optimization of Double Square FSS for Sub 6G Applications Using ML Algorithms

Frequency Selective Surfaces (FSS) play a very critical role now-a-days in electromagnetic engineering, which enables precise control of electromagnetic wave propagation for various applications. This article explores the integration of Machine Learning (ML) algorithms to optimize the structure to achieve enhanced performance and efficiency for Sub 6G applications (6 GHz and 2.4 GHz). The Double Square Frequency Selective Surface (DS-FSS) consists of two distinct layers of square shaped periodic patterns. The substrate used is RT Duroid 5880 which has a dielectric constant of 2.2 with a width of 1.6 mm for better characteristics. The periodicity of the unit cell is 20 mm x 20 mm which makes it isotropic and polarization insensitive. This configuration provides greater flexibility in tailoring the overall frequency response compared to single layer FSS. A comprehensive data set (approx. 5000) has been prepared by simulating the electromagnetic responses from the Equivalent Circuit Model (ECM) which are fed to the ML algorithm to extract the features and build the regression model and to estimate the optimal parameters. The ML algorithms used are Linear Regression and Random Forest. These are then validated against the electromagnetic simulations using HFSS Software. Two stop bands, one at 2.4 GHz and the other at 6.3 GHz and one pass band at 3.8 GHz has been achieved. Shielding Effectiveness of about 49 dB is achieved at 2.4 GHz and about 57 dB is achieved at 6.3 GHz. The angular stability of 0° to 80°, with a stable bandwidth, is also achieved by the proposed design which is better than the existing works.