摘要
提出了一种基于卷积神经网络技术的非多普勒激光雷达测风系统。该系统利用半导体感光元件(charge coupled device,CCD)拍摄气溶胶颗粒物的激光雷达后向散射图,针对不同风速气溶胶颗粒物的运动轨迹特征,实现定量的风速测量。卷积神经网络对经过图像预处理的气溶胶颗粒物运动轨迹图,进行特征提取并生成风速测量模型,第100次训练样本和验证样本的准确率分别为0.92和0.93。利用生成的风速测量模型对测试样本进行实验,准确率达到0.84。这种低成本、操作简便的非多普勒激光雷达测风系统,能够解决当前多普勒频移测风激光雷达成本高的痛点,具有很强的现实意义。
A Non-Doppler lidar wind measurement system based on convolutional neural network technology is proposed.The system uses(charge coupled device,CCD)to capture the lidar backscatter map of aerosol particles,and realizes quantitative wind speed measurement according to the movement trajectory of aerosol particles at different wind speeds.The convolutional neural network extracts the features of the aerosol particle movement trajectory after lidar image preprocessing and generates a wind speed measurement model.The accuracy of the 100 th training sample and verification sample are 0.92 and 0.93,respectively.The wind speed measurement model is used on the test sample,and the accuracy rate reaches 0.84.This low-cost,easy-to-operate non-Doppler lidar wind measurement system can effectively reduce the cost of wind measurement and has strong practical significance.
作者
张平慧
胡淼
许蒙蒙
应娜
贺文迪
金益文
赵喻晓
周雪芳
杨国伟
毕美华
李齐良
ZHANG Pinghui;HU Miao;XU Mengmeng;YING Na;HE Wengdi;JIN Yiwen;ZHAO Yuxiao;ZHOU Xuefang;YANG Guowei;BI Meihua;LI Qiliang(College of Communication Engineering,Hangzhou Dianzi University,Hangzhou,Zhejiang 310018,China;State Key laboratory of NBC Protection for Civilian,Beijing 102205,China)
出处
《光电子.激光》
CAS
CSCD
北大核心
2021年第10期1039-1045,共7页
Journal of Optoelectronics·Laser
基金
国家自然科学基金(61705055)
2020年度浙江省重点研发计划项目(2019C01G1121168)
2020年国家级大学生创新创业训练项目(202010336054)
2020年浙江省大学生科技创新活动计划暨新苗人才计划(2020R407073)
杭州电子科技大学研究生科研创新基金资助项目。