基于TDLAS技术的天然气中痕量硫化氢分析的PLS算法应用

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.

Partial Least Squares Algorithm Application in TDLAS Based Trace H2S Analyses in Natural Gas

When the Tunable Diode Laser Absorption Spectroscopy (TDLAS) analyzer is used to measure the hydrogen sulfide in nature gas,there are strong interferences from light hydrocarbon and CO2 in the natural gas background.Their spectra structure is imposed on the analyze spectra.The spectral interferences affect the extraction of the H2S absorption spectrum.The Partial Least Squares(PLS) algorithm is proposed to deal with the spectral interferences.The Common Least Square (CLS) and the PLS models are adopted in analyzer to evaluate the differential spectrum and direct measurement spectrum of 0～50 ppm H2S in natural gas.PLS is superior than CLS in the analysis of both differential spectrum and direct spectrum.Based the analyzer reading,the absolute error of PLS result with direct measurement spectrum is within ±1 ppm.With PLS algorithm,the analyzer need not to store the mass data of reference spectra,and the complicated differential system can be elimentated from analyzer,which can save the cost and improve robustness and the time response.