采用BP神经网络补偿的激光气体检测系统研制

Development of laser gas detection system with BP neural network compensation

海底天然气开采是目前我国实行多元化能源结构的重要一环,而组成天然气的成分包含有一氧化碳和甲烷,实现对一氧化碳的检测在深海天然气开采具有积极意义。因此本文基于近红外VCSEL激光器实现了适应性较好的痕量一氧化碳气体检测系统。为适应不同气体检测环境的需要,本文依据反演模型随温度压强变化的特点设计了对应的补偿算法。考虑到采用二元二次函数拟合气体温度压强补偿系数的误差较大,本文基于BP神经网络实现了更加准确的气体温度压强补偿算法,为本系统在深海天然气开采探测方面的应用奠定了基础。

Deep sea natural gas exploitation is an important part of China’s diversified energy structure,and the components of natural gas include carbon monoxide and methane.The detection of carbon monoxide is of positive significance in deep-sea natural gas exploitation.Therefore,a trace carbon monoxide detection system with good adaptability is realized based on near infrared VCSEL laser.In order to meet the needs of different gas detection environments,this paper designs the corresponding compensation algorithm according to the characteristics of the inversion model varying with temperature and pressure.The error of fitting gas temperature and pressure compensation coefficient with binary quadratic function is large.In this paper,a more accurate gas temperature and pressure compensation algorithm is realized based on BP neural network,which greatly improves the possibility of applying laser gas detection system to deep-sea natural gas exploitation.