光学学报, 2019, 39 (8): 0801002, 网络出版: 2019-08-07
基于卷积神经网络的大气中光路气流扰动实验研究 下载: 1370次
Atmospheric Optical Path Airflow Disturbance Analysis Method Based on Convolutional Neural Network
大气光学 空间光学 气流扰动 卷积神经网络 深度学习 atmospheric optics free space optics air flow disturbance convolution neural network deep learning
摘要
提出了一种基于激光光斑畸变和卷积神经网络(CNN)的光路气流扰动研究方案。利用CNN对激光光束在空间传播中受到气流扰动后的光斑畸变进行学习,得到光束传播路径上的气流扰动情况。实验表明,训练得到的评估参数与由风速仪测得的光路中的气流扰动(风速)具有强相关性。本方案提供了一种短距离、快速、低成本的气流扰动分析手段。
Abstract
A method to investigate optical path turbulence based on laser spot distortion and a convolutional neural network (CNN) is proposed. Utilizing the CNN, we evaluated the spot distortion of laser beams resulting from airflow disturbance in space propagation. As a result, details of turbulence on the beam propagation path can be obtained. Experimental results demonstrate a high correlation between the evaluation parameter and the turbulent intensity (wind speed) measured by an anemoscope. The proposed method provides a turbulence analysis with short distance, high speed, and low cost.
刘一琛, 吴侃, 邱高峰, 陈建平. 基于卷积神经网络的大气中光路气流扰动实验研究[J]. 光学学报, 2019, 39(8): 0801002. Yichen Liu, Kan Wu, Gaofeng Qiu, Jianping Chen. Atmospheric Optical Path Airflow Disturbance Analysis Method Based on Convolutional Neural Network[J]. Acta Optica Sinica, 2019, 39(8): 0801002.