An Inverse Topological Design Method (ITDM) Based on Machine Learning for Frequency-Selective-Surface (FSS) Structures

To efficiently and conveniently realize the topology design with many degrees of freedom (DoFs), this work proposes an inverse topological design method (ITDM) based on machine learning for frequency-selective-surface (FSS) structures from a novel perspective. In the proposed ITDM, the input is a set of desired
$\vert \text{S}11\vert $
curves, and the output is the predicted FSS structure presented as an image to directly reveal the distribution of binary pixels in the modeling domain. To obtain the training dataset, an efficient training dataset construction strategy is proposed based on several classical FSS structures. The advantage of this approach is that the training samples can be guaranteed with high-quality; thus, the convergence of the model is faster. Two numerical examples of single-layer and double-layer FSS structures involving large numbers of binary variables (900 and 1800, respectively) are employed to validate the effectiveness of the proposed ITDM. Meanwhile, the fabricated topological structures are also measured to verify the performance of a designed FSS obtained from ITDM.