Efficient spectrum prediction and inverse design for plasmonic waveguide systems based on artificial neural networks

Abstract

In this article, we propose a novel approach to achieve spectrum prediction,\nparameter fitting, inverse design and performance optimization for the\nplasmonic waveguide coupled with cavities structure (PWCCS) based on artificial\nneural networks (ANNs). The Fano resonance and plasmon induced transparency\neffect originated from the PWCCS have been selected as illustrations to verify\nthe effectiveness of ANNs. We use the genetic algorithm to design the network\narchitecture and select the hyper-parameters for ANNs. Once ANNs are trained by\nusing a small sampling of the data generated by Monte Carlo method, the\ntransmission spectrums predicted by the ANNs are quite approximate to the\nsimulated results. The physical mechanisms behind the phenomena are discussed\ntheoretically, and the uncertain parameters in the theoretical models are\nfitted by utilizing the trained ANNs. More importantly, our results demonstrate\nthat this model-driven method not only realizes the inverse design of the PWCCS\nwith high precision but also optimizes some critical performance metrics for\ntransmission spectrum. Compared with previous works, we construct a novel\nmodel-driven analysis method for the PWCCS which are expected to have\nsignificant applications in the device design, performance optimization,\nvariability analysis, defect detection, theoretical modeling, optical\ninterconnects and so on.\n