基于深度学习的Fano共振超材料设计

Fano resonances design of metamaterials based on deep learning

本文提出了一种基于深度学习的超材料Fano共振设计方法,能够获得高Q共振的线宽、振幅和光谱位置特性。利用深度神经网络建立结构参数和透射谱曲线之间的映射,正向网络实现对透射谱的预测,逆向网络实现对高Q共振按需设计,设计过程中实现了低均方误差(MSE),训练集的均方误差为0.007。与传统方法需要耗时的逐个数值模拟相比,深度学习设计方法大大简化了设计过程,实现了高效、快速的设计目标。对Fano共振的设计也可推广应用到其它类型的超材料的自动逆向设计,显著提高了更复杂的超材料设计的可行性。

In this paper,a metamaterial Fano resonance design method based on deep learning is proposed to obtain high-quality factor(high-Q)resonances with desired characteristics,such as linewidth,amplitude,and spectral position.The deep neural network is used to establish the mapping between the structural parameters and the transmission spectrum curve.In the design,the forward network is used to predict the transmission spectrum,and the inverse network is used to achieve the on-demand design of high Q resonance.The low mean square error(MSE)is achieved in the design process,and the mean square error of the training set is 0.007.The results indicate that compared with the traditional design process,using deep learning to guide the design can achieve faster,more accurate,and more convenient purposes.The design of Fano resonance can also be extended to the automatic inverse design of other types of metamaterials,significantly improving the feasibility of more complex metamaterial designs.