摘要
以CO_(2)为主的温室气体排放使得全球变暖,严重影响生态环境,2021年习近平主席在二十国集团领导人峰会上提出“中国将力争2030年前实现碳达峰、2060年前实现碳中和”,因此精确检测CO_(2)气体浓度具有重要研究意义。由于CO_(2)气体吸收谱线的展宽受到气体压力、温度等因素影响,导致TDLAS气体检测系统测量结果误差增大,因此本文结合HITRAN数据库仿真,提出了基于BP神经网络深度学习的CO_(2)浓度反演算法和嵌入式实现方法,实现了对气体浓度的补偿,为嵌入式浓度反演算法设计提供理论依据。该算法可以移植到STM32F407中,经过测试,气体浓度的检测误差小,有效提升了气体检测精度,此方法同样适用于TDLAS型的其他气体检测应用场景中。
The CO_(2)Greenhouse gas emissions cause global warming and seriously affect the ecological environ-ment.In 2021,The President Xi proposed at the G20 Summit that"China will strive to achieve carbon peak by 2030 and carbon neutrality by 2060".Therefore,accurate detection of CO_(2)concentration has important research signifi-cance.Due to the influence of factors such as gas pressure and temperature on the broadening of CO_(2)gas absorption spectral lines,the measurement error of TDLAS gas detection system increases.Therefore,this paper proposes a CO_(2)concentration inversion algorithm and embedded implementation method based on deep learning of BP neural network,combined with HITRAN database simulation,to achieve compensation for gas concentration and provide a theoretical basis for the design of embedded concentration inversion algorithms.This algorithm can be ported to STM32F407.Af-ter testing,the detection error of gas concentration is small,effectively improving gas detection accuracy,and this method is also applicable to other gas detection application scenarios of TDLAS type.
作者
王彪
杨子腾
卞广雨
王冠懿
赵奕飞
薛金波
程林祥
WANG Biao;YANG Ziteng;BIAN Guangyu;WANG Guanyi;ZHAO Yifei;XUE Jinbo;CHENG Linxiang(Changchun Institute of Optics,Fine Mechanics and Physics,Chinese Academy of Sciences,Changchun 130033,China;Jilin University,Changchun 130015,China;Jilin Agricultural University,Changchun 130118,China;University of Chinese Academy of Sciences,Beijing 100049,China)
出处
《激光杂志》
CAS
北大核心
2023年第5期42-46,共5页
Laser Journal
基金
吉林省科技发展计划重点科技研发项目(No.20220203016SF)
中国科学院大学生创新实践训练计划项目(No.2022008090)。
