A Conformal Frequency Selective Rasorber With Intelligent Band-Transfer Abilities via Jointly Learning Network

Abstract

Frequency selective rasorber (FSR) has been applied in many fields due to the specific ability to filter and absorb electromagnetic waves by designing the geometrical shapes and the structural parameters elaborately. However, the bulky structure and low angular stability of conventional FSR conflict with the conformal application. Further, the response of FSR is determined by multiparameters, making the design process based on time-consuming simulation and empirical turning. In this article, a kind of conformal FSR with high angular stability is proposed, where an inverse design method called a jointly learning network with low input dimension and high accuracy is further introduced to intelligently tune the FSR to the other band. Under normal incidence, the proposed FSR has a passband at 6.64 GHz with an insertion loss of 0.05 dB and has two absorption bands with an absorption rate above 80% in the ranges of 3.26–5.73 GHz and 7.67–10.28 GHz. Further, the angular stability is verified under oblique angles up to 60°. The consistency between the simulated and measured results shows the feasibility of its conformal applications. Subsequently, the results of the structures predicted by the deep learning method demonstrate the strong generalization ability of the proposed FSR.