基于人工神经网络的低非线性空芯反谐振光纤研究
Research on Low-Nonlinearity Hollow-Core Anti-Resonant Fiber Based on Artificial Neural Network
摘要
本文针对传统石英光纤在高速率、大容量光通信中面临的材料色散、非线性效应等瓶颈问题，提出了一种基于深度学习的空芯反谐振光纤（HC-ARF）结构优化设计方法。通过融合正交试验设计（DOE）与人工神经网络（ANN），构建了空间映射神经网络（SMNN）代理模型，以高效探索高维参数空间并建立结构参数与光学性能之间的非线性映射关系。该方法利用粗网格仿真生成低成本训练数据，结合少量高精度细网格仿真进行校准，显著降低了有限元仿真的计算成本与时间。优化结果表明，所提方法在保证设计精度的同时，较传统遗传算法和神经网络优化方法显著提升了收敛速度与计算效率，为低损耗、低非线性、高模场面积的空芯反谐振光纤设计提供了系统化解决方案，对光通信、能源传输及光电集成技术的发展具有重要参考价值。
Abstract
This paper addresses the bottlenecks of material dispersion and nonlinear effects faced by traditional quartz optical fibers in high-speed, high-capacity optical communication systems. A deep learning-based structural optimization design method for hollow-core anti-resonant fiber (HC-ARF) is proposed. By integrating the Design of Experiments (DOE) and an Artificial Neural Network (ANN), a Space-Mapping Neural Network (SMNN) surrogate model is constructed to efficiently explore the high-dimensional parameter space and establish a nonlinear mapping relationship between structural parameters and optical performance. This method utilizes coarse-mesh simulations to generate low-cost training data, which is then calibrated with a small amount of high-fidelity fine-mesh simulations, significantly reducing the computational cost and time associated with finite element simulations. The optimization results demonstrate that the proposed method achieves a substantial improvement in convergence speed and computational efficiency while maintaining design accuracy, compared to traditional genetic algorithms and conventional neural network optimization methods. It provides a systematic solution for designing HC-ARFs with low loss, low nonlinearity, and large mode area, offering significant reference value for the advancement of optical communications, energy transmission, and photonic integration technologies.关键词
空芯反谐振光纤/结构优化/人工神经网络ANN/正交试验设计/有限元仿真/非线性效应Key words
Hollow-core anti-resonant fiber/structural optimization/artificial neural network (ANN)/orthogonal experimental design/finite element simulation/nonlinear effect引用本文复制引用
陈方楷,张治国.基于人工神经网络的低非线性空芯反谐振光纤研究[EB/OL].(2026-02-03)[2026-02-05].http://www.paper.edu.cn/releasepaper/content/202602-18.学科分类
光电子技术
评论