基于波前预测模型的波前校正方法

马玉磊; 杜广鑫

您当前的位置：

首页 >

文章列表页 >

基于波前预测模型的波前校正方法

光信息技术 | 更新时间：2026-03-24

- 基于波前预测模型的波前校正方法
- Wavefront correction method based on wavefront prediction model
- 光学技术 2026年52卷第2期页码：203-210
- 作者机构：
  
  1.新乡学院计算机与信息工程学院，河南新乡 453000
  2.中国人民解放军32096部队，湖北武汉 430074
- 作者简介：
  
  马玉磊（1982— ），男，副教授，从事计算机网络研究。mayulei8219@163.com
  杜广鑫（1996— ），男，助理工程师，从事大数据分析与信息传播研究。
- 基金信息：
  
  河南省科技厅重点研发与推广专项（科技攻关）项目(212102210405);新乡学院教育教学改革研究与实践项目成果（31）
- DOI：
  中图分类号： TP394.1;TH691.9
- 收稿：2025-12-01，
  
  修回：2026-01-09，
  
  纸质出版：2026-03-20
- 稿件说明：
移动端阅览
马玉磊,杜广鑫.基于波前预测模型的波前校正方法[J].光学技术,2026,52(02):203-210.

MA Yulei,DU Guangxin.Wavefront correction method based on wavefront prediction model[J].OPTICAL TECHNIQUE,2026,52(02):203-210.
马玉磊,杜广鑫.基于波前预测模型的波前校正方法[J].光学技术,2026,52(02):203-210. DOI：

MA Yulei,DU Guangxin.Wavefront correction method based on wavefront prediction model[J].OPTICAL TECHNIQUE,2026,52(02):203-210. DOI：

摘要

针对自适应光学中波前传感器与波前校正之间的时延问题，提出一种基于波前预测模型的自适应光学波前实时校正方法。波前预测模型由两个并行分支构成，第一个分支基于波前时间序列计算差分波前序列以得到大气湍流的时域动态演化信息，将其传入长短期记忆网络提取大气湍流演化的动态特征；第二个分支将波前时间序列传入增强的Transformer编码器提取波前序列的时间域依赖性，并使用多尺度卷积层提取每个波前系数的空间域特征

从而得到包含畸变波前时空域信息的特征。最终，通过交叉注意力机制动态地从第一个分支输出特征中筛选与当前波前预测最相关的动态特征，基于动态特征与Transformer编码特征的融合信息进行波前预测。仿真结果表明，在强、中与弱湍流下，所提波前预测模型的均方根误差分别比次优模型减少43.8%、43.9%和57.9%。

Abstract

Aiming at the problem of time delay between wavefront sensor and wavefront correction in adaptive optics

a wavefront real-time correction method based on wavefront prediction model for adaptive optics is proposed. The wavefront prediction model consists of two parallel branches

in the first branch

the difference wavefront sequence is computed based on wavefront time sequence to generate the temporal dynamic evolutionary information of the atmospheric turbulence

and it is delivered to long short term memory network to extract the dynamic features of atmospheric turbulence evolution. In the second branch

the wavefront time sequence is delivered to the enhanced Transformer encoder to extract the temporal dependency of the time sequence

and a multi-scale convolutional neural layer is introduced to extract the spatial features of each wavefront Zernike coefficient

to generate features including temporal and spatial information of the aberration wavefronts. Finally

the dynamic features most relevant to current wavefront prediction are selected from the features come from the first branch by using of cross-attention mechanism

and the future wavefront are predicted based on the information with combination of dynamic features and Transformer encoded features. Simulation results show that under the conditions of strong

moderate and weak turbulent atmospheric

the root mean square error of the proposed wavefront prediction model decreases by 43.8%

43.9% and 57.9% respectively compared to the second-best method.

关键词

Keywords

references

刘莹奇 , 陆红强 , 舒营恩 , 等 . 机载战术激光武器的低阶自适应光学校正技术研究 [J ] . 应用光学 , 2025 , 46 （ 2 ）: 412 — 418 .

Liu Yingqi , Lu Hongqiang , Shu Yingen , et al . Research on correction technology of low order adaptive optics for airborne tactical laser weapons [J ] . Journal of Applied Optics , 2025 , 46 （ 2 ）: 412 — 418 .

Pantong Y , Rui L , David M K . Construction and use of an adaptive optics two-photon microscope with direct wavefront sensing [J ] . Nature Protocols Erecipes for Researchers , 2023 , 18 （ 12 ）: 3732 — 3766 .

Li J , Zhang Z , Gao J , et al . Bandwidth of adaptive optics system in atmospheric coherent laser communication [J ] . Optics Communications , 2016 , 359 （ 7 ）: 254 — 260 .

黄盛炀 , 习锋杰 , 宁禹 , 等 . 相位光栅型曲率传感器的波前校正算法 [J ] . 强激光与粒子束 , 2011 , 23 （ 10 ）: 2573 — 2577 .

Huang Shengyang , Xi Fengjie , Ning Yu , et al . Reconstruction algorithm for wavefront curvature sensor with phase grating [J ] . High Power Laser and Particle Beams , 2011 , 23 （ 10 ）: 2573 — 2577 .

卞点 , 骆雅婷 , 侯登峰 , 等 . 偏振可调谐式剪切干涉的大畸变波前实时重构 [J ] . 光学精密工程 , 2024 , 32 （ 15 ）: 2325 — 2333 .

Bian Dian , Luo Yating , Hou Dengfeng , et al . Polarization-tunable shearing interferometry method for real-time reconstruction of large aberration wavefronts [J ] . Optics and Precision Engineering , 2024 , 32 （ 15 ）: 2325 — 2333 .

梁静远 , 庞明志 , 柯熙政 . 无线光通信中大气湍流抑制方法 [J ] . 电子测量与仪器学报 , 2024 , 38 （ 11 ）: 1 — 14 .

Liang Jingyuan , Pang Mingzhi , Ke Xizheng . Atmospheric turbulence suppression method in optical wireless communication [J ] . Journal of Electronic Measurement and Instrument , 2024 , 38 （ 11 ）: 1 — 14 .

Caroline Kulcsár , Raynaud H F , Petit C , et al . Minimum variance prediction and control for adaptive optics [J ] . Automatica , 2012 , 48 （ 9 ）: 1939 — 1954 .

Poyneer L A , Jean-Pierre Véran . Kalman filtering to suppress spurious signals in adaptive optics control [J ] . Journal of the Optical Society of America A Optics Image Science & Vision , 2010 , 27 （ 11 ）: 223 — 234 .

Gray M , Petit C , Rodionov S , et al . Local ensemble transform Kalman filter, a fast non-stationary control law for adaptive optics on ELTs: theoretical aspects and first simulation results [J ] . Optics Express , 2014 , 22 （ 17 ）: 894 — 913 .

Cerqueira P , Piscaer P , Verhaegen M . Sparse data-driven wavefront prediction for large-scale adaptive optics [J ] . Journal of the Optical Society of America .A,Optics,image science, and vision, 2021 , 38 （ 7 ）: 992 — 1002 .

Zhu L , Gao Z , Ge X , et al . Highly stable spatio-temporal prediction network of wavefront sensor slopes in adaptive optics [J ] . Sensors , 2023 , 23 （ 22 ）: 9260 — 9260 .

Xu Zhenxing , Yang Ping , Hu Ke , et al . Deep learning control model for adaptive optics systems [J ] . Applied Optics , 2019 , 58 （ 8 ）: 1998 — 2009 .

邱从攀 , 刘国栋 , 张大勇 , 等 . 深度学习在波前重构与波前预测中的研究进展 [J ] . 强激光与粒子束 , 2024 , 36 （ 7 ）: 071002.1 — 071002.10 .

Qiu Congpan , Liu Guodong , Zhang Dayong , et al . Research progress in deep learning for wavefront reconstruction and wavefront prediction [J ] . High Power Laser and Particle Beams , 2024 , 36 （ 7 ）: 071002.1 — 071002.10 .

Wang N , Zhu L , Ma S , et al . Deep learning-based prediction algorithm on atmospheric turbulence-induced wavefront for adaptive optics [J ] . Photonics Journal ,IEEE, 2022 , 14 （ 5 ）: 1 — 10 .

Chen J , Shah V , Liu L . Performance of a U-Net-based neural network for predictive adaptive optics [J ] . Optics Letters , 2021 , 46 （ 10 ）: 2513 — 2516 .

Xuewen L , Tim M , Chris S , et al . Wavefront prediction using artificial neural networks for open-loop adaptive optics [J ] . Monthly Notices of the Royal Astronomical Society , 2020 , 496 （ 1 ）: 456 — 464 .

Chen Y . LSTM recurrent neural network prediction algorithm based on Zernike modal coefficients [J ] . Journal for Light-and Electronoptic , 2020 , 203 （ 9 ）: 163796.1 — 163796.8 .

Wu J , Tang J , Zhang M , et al . PredictionNet: a long short-term memory-based attention network for atmospheric turbulence prediction in adaptive optics [J ] . Applied Optics , 2022 , 61 （ 13 ）: 3687 — 3694 .

闫丽娟 , 梁冰 . 大气湍流下激光通信的波前还原研究 [J ] . 光学技术 , 2025 , 51 （ 2 ）: 210 — 217 .

Yan Lijuan, Liang Bin, Study on wavefront recovery of laser communication over atmospheric turbulence [J ] . Optical Technique , 2025 , 51 （ 2 ）: 210 — 217 .

Dai G M . Modal compensation of atmospheric turbulence with the use of Zernike polynomials and Karhunen–Loève functions [J ] . Journal of the Optical Society of America A , 1995 , 12 （ 10 ）: 2182 — 2193 .

Vaswani A , Shazeer N , Parmar N , et al . Attention is all you need [C ] // Proceedings of the 31st International Conference on Neural Information Processing Systems（NIPS 2017 . LongBeach, USA, IEEE : 2017 : 5998 — 6008 .

王宁 , 朱里程 , 葛欣兰 , 等 . 基于注意力机制的自适应光学开环斜率预测 [J ] . 中国激光 , 2024 , 51 （ 6 ）: 0605001.1 — 0605001.10 .

Wang Ning , Zhu Licheng , Ge Xinlan , et al . Open-loop slope prediction in adaptive optics based on attention mechanism [J ] . Chinese Journal of Lasers , 2024 , 51 （ 6 ）: 0605001.1 — 0605001.10

Guyon E , Wesfreid J E , Roux S . Experiments with rayleigh-bénard convection [M ] . Dynamics of Spatio-Temporal Cellular Structures. Berlin, Heidelberg : Springer , 2007 : 67 — 94 .

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于多步匹配的超声多目标跟踪方法与系统构建

基于残差注意力机制的离轴数字全息重建

基于有限内存拟牛顿法的荧光显微成像相位差异自适应光学技术研究

基于深度学习与FPGA的涡旋光束OAM模式识别

长距离光纤通信系统的非线性损伤补偿方法