Study on the choices of design parameters for inverse design of metasurface using Deep leargning

Metasurfaces with fascinating electromagnetic property have already achieved wide applications in academia and industry. The traditional design approach for metasurfaces highly relies on full-wave numerical simulations and trial-and-error methods. It is time-consuming and laborious to obtain the optimal design parameters. Recently, extensive researches have shown advantages and superiority of the deep learning method in solving non-intuitive problem. Several attempts have been made to demonstrate Artificial Intelligence (AI) usage in the electromagnetic field. In this paper, a deep-learning-based method has been proposed and demonstrated numerically. Unlike previous deep-learning-based design methods for metasurfaces which directly use the physical geometry structure parameters to predict the metasurface's response, this method leverage the electric size to predict the response of the metasurface. Compared the method that using physical structure parameters, our method is more accurate and it also needs less training data to reach acceptable results.