Neural Network Modeling for the Construction of Flexible Frequency Selective Surfaces

Scattering analysis for flexible Frequency Selective Surfaces (FSSs) is important. The infinitesimal dipole model (IDM) method can achieve high computational efficiency in scattering analysis. However, the complex preprocessing essentially limits the popularization and application of this method in large-scale computational problems. In this letter, an innovative method based on the IDM and backpropagation neural network (BPNN) is proposed to address the limitations of flexible FSS. First, the constrained infinitesimal dipole (ID) is solved by singular value decomposition (SVD), and the unit information in the FSS is restored by model reconstruction. Afterward, the BPNN was used to learn the dipole generation process, further improving efficiency. The simulation and measurement results demonstrate that the proposed method yields results comparable to those of the traditional method in terms of calculation accuracy. This letter presents a novel solution to the problem of FSSs in electromagnetic simulations and demonstrates the potential of deep learning technology in electromagnetism.