基于TDLAS技术的天然气中痕量硫化氢分析的PLS算法应用

发布时间：2018-12-13 21:00

【摘要】：基于可调谐二极管激光吸收光谱技术的光谱分析仪测量天然气中硫化氢浓度的过程中,各种轻烃和二氧化碳等背景成分对光谱产生干扰,导致多个吸收峰叠加,对提取吸收光谱真实特征造成影响.应用偏最小二乘法消除背景成分的干扰,设计了检测天然气中硫化氢气体的可调谐二极管激光吸收光谱实验系统,采用偏最小二乘法和最小二乘法模型,分别检测了天然气中0~50ppm硫化氢成分的直测光谱和差分光谱.偏最小二乘法算法的测量结果均优于最小二乘法算法,且偏最小二乘法算法对于直测谱的测量误差保持在±1ppm范围内,满足分析仪器2%的准确度要求.利用偏最小二乘法算法避免了最小二乘法所必须的大量参考光谱数据的存储,分析仪可省掉复杂的差分光谱系统,从而达到降低成本、提高系统鲁棒性和实时性的效果.
[Abstract]:In the process of measuring the concentration of hydrogen sulfide in natural gas by a spectrum analyzer based on tunable diode laser absorption spectroscopy, various background components, such as light hydrocarbon and carbon dioxide, interfere with the spectrum, resulting in the superposition of multiple absorption peaks. It has an effect on extracting the true characteristics of absorption spectrum. The experiment system of tunable diode laser absorption spectrum for detecting hydrogen sulfide in natural gas was designed by using partial least square method to eliminate the interference of background components. The partial least square method and the least square method model were used. The direct and differential spectra of 0~50ppm hydrogen sulfide in natural gas were measured. The results of partial least square algorithm are better than that of least squares algorithm, and the measurement error of partial least squares algorithm for direct measurement spectrum is kept in the range of 卤1ppm, which meets the requirement of 2% accuracy of analytical instrument. The partial least squares algorithm is used to avoid the storage of a large number of reference spectral data required by the least square method, and the complex differential spectrum system can be saved by the analyzer, which can reduce the cost and improve the robustness and real-time performance of the system.
【作者单位】：中国石油大学(华东)信息与控制工程学院;
【基金】：华东师范大学精密光谱科学与技术国家重点实验室开放课题(No.F1505037A) 中国石油大学(华东)自主创新项目(Nos.27R1505012A,27R1505024A);中国石油大学(华东)研究生创新工程资助项目(No.YCXJ2016076)资助~~
【分类号】：TE642;TN249

【相似文献】