Surrogate-based model for metamaterial synthesis

Abstract

In this work, we propose a novel computer-aided design (CAD) model design for metamaterial analysis, based on neuro-space mapping. Neuro-space mapping is a combination of two robust optimisation techniques, artificial neural network (ANN) and space mapping (surrogate-based modelling). The developed model calculates the permeability of a metamaterial unit cell, for which closed-form formula is not available. The model takes the dimensions of the metamaterial unit cell as the input, and then gives the output as the split ring resonator (SRR) dimension. From the SRR dimension, the material parameters are calculated using available closed-form formula. The model is first developed with a Ω-atom unit cell, and then verified applying the model to a deformed-Ω atom unit cell.