Design of Optically Transparent Coded Metamaterial Based on an Indium Tin Oxide Film Using Deep Learning for Radar Cross-Section Reduction
Zhihui Wang, Hui Luo, Yongzhi Cheng, Fu Chen, Xiangcheng Li
Abstract
A multimechanism optically transparent coded metamaterial absorber is designed and fabricated to suppress backward scattering based on deep neural network (DNN) and particle swarm algorithm (PSO). The DNN is utilized for the rapid prediction of the amplitude and phase and simplifies the design process. The PSO facilitates global exploration and automatic calibration of the amplitude and phase settings to achieve an optimal structural configuration. The integration of DNN and PSO results in an innovative inverse design algorithm, enhancing efficiency and automation in multimechanism metasurface absorber (MMA) design. The designed coding metamaterial structure enables the realization of amplitude modulation and phase cancellation of electromagnetic waves, achieving significantly reduced backward scattering within 7.5–17 GHz at a total thickness of only 2.5 mm. The loss mechanisms were explored by electric field, magnetic field, power loss density, and surface current distributions. The methodology not only offers a perspective on RCS reduction but also demonstrates versatility for designing other metamaterial-based functional devices.