Predicting complete band-gaps of 2D photonic crystals by using artificial neural networks

摘要

In this paper, artificial neural networks are modeled to predict complete band-gaps of bi-dimensional photonic crystals. The available data-set has been generated by an integrated artificial immune network and MPB (MIT Photonic Bands) optimization procedure. Two case studies were carried out, considering square lattice photonic crystals composed of two and three silicon round rods embedded in air. Results from tests showed the modeled artificial neural networks are capable of estimating complete band-gaps across the proposed range of rods.