基于卷积神经网络的大气中光路气流扰动实验研究

Atmospheric Optical Path Airflow Disturbance Analysis Method Based on Convolutional Neural Network

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。

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.