Design of resonant metasurface absorber using feed-forward neural network

In this work, an ultrathin (lambda 0/50,lambda 0=122.45 ${\lambda }_{0}\unicode{x02215}50,{\lambda }_{0}=122.45$ mm) design of a frequency-selective metamaterial absorber (MTA) is proposed. An absorption of 100% is achieved at the resonant frequency fr=2.45 ${f}_{{\rm{r}}}=2.45$ GHz with a fractional bandwidth of 1.45%. The MTA unit cell geometry is parametrized and modeled in a full-wave electromagnetic simulator. Using a data set generated by varying the absorber unit cell geometry and simulating it in SIMULIA CST Studio Suite, the feed-forward neural network is able to learn the relationship between the physical structure and its electromagnetic response. The numerical results for the MTA performance have been confirmed by an experiment. An array of 25 MTA elements in a 5x5 $5\times 5$ configuration has been fabricated and tested in an anechoic chamber. The simulated and measured results are in a good agreement.