Deep Learning and ECM Fusion for Realization of Advanced Resistive Ink FSS-Based Customized Microwave Absorber

Abstract

This letter introduces a simple frequency-selective surface (FSS) design for a resistive microwave absorber in X-band applications. It utilizes a polymer-based dielectric substrate and Y-Shield HSF 64 resistive ink with a conductivity of 640 S/m. The absorber’s design is realized using a deep neural network (DNN) integrated with equivalent circuit modeling (ECM) and validated through the bees algorithm (BeA). Fabrication involves manually creating $1\times 3$ unit cells, followed by experimental evaluation using the WR-90 rectangular waveguide method. Results from ECM-backed DNN, ECM-assisted BeA, and full-wave simulation align closely with measured data, demonstrating a minimum reflection coefficient of −22.5 dB at the resonant frequency of 10 GHz with a bandwidth of 4.2 GHz (8.2–12.4 GHz) for normal incidence. This letter also examines the surface current distribution and electromagnetic (EM) properties, highlighting the absorber’s simplicity, flexibility, lightweight construction, polarization insensitivity, angular stability, and wideband characteristics, making it suitable for customized stealth and electromagnetic- compatibility (EMC) applications.