Complete band-gap prediction of 2D photonic crystals by using multilayer perceptron

Abstract

In this paper, a Multilayer Perceptron artificial neural network is modeled to estimate complete photonic band-gaps (C-PBGs) of bi-dimensional photonic crystals. Unit cells of square lattice photonic crystals, composed of two silicon round rods and embedded in air, have been designed by an artificial immune network algorithm, and their geometries have been stored in a database along with their C-PGBs. Tests using unknown unit cells assess and demonstrate the C-PBG predicting capability of the modeled Multilayer Perceptron.