本文選題：LASSO + FTIR��； 參考：《光譜學與光譜分析》2017年10期

【摘要】：采用紅外光譜技術對未知氣體組分進行監(jiān)測,需要對氣體組分進行定性識別分析�；诙嘣�性回歸模型的LASSO變量選擇技術廣泛應用于數(shù)據(jù)分析領域。將LASSO方法引入到紅外光譜分析領域,提出一種LASSO變量選擇技術結合循環(huán)線性最小二乘(LCLS)分析的定性識別方法,并開展了相關的實驗對其進行驗證。實驗采集CO,C_2H_4,NH_3,C_3H_8,C_4H_(10)和C_6H_(14)六種單組分傅里葉變換紅外(FTIR)光譜吸光度譜以及一組C_2H_4和NH_3混合組分的吸光度譜,結合實驗室自建光譜數(shù)據(jù)庫,先采用LASSO方法對采集的光譜進行初步定性分析,然后使用LCLS方法剔除干擾組分。實驗結果表明,LASSO結合LCLS的方法能有效識別出光譜中的目標組分,即使是在干擾嚴重的光譜波段也可以剔除掉大部分的干擾組分。
[Abstract]:Infrared spectroscopy is used to monitor unknown gas components, which need to be identified qualitatively. LASSO variable selection technology based on multivariate linear regression model is widely used in data analysis field. In this paper, the LASSO method is introduced into the field of infrared spectrum analysis, and a qualitative identification method based on LASSO variable selection technique and cyclic linear least squares LCLS analysis is proposed, and the relevant experiments are carried out to verify it. Experimental collection of the absorbance spectra of the six kinds of one-component Fourier transform infrared (FTIRIRs) and a group of C _ 2H _ 4 and NH _ 3 mixed components, and a laboratory self-built spectral database, a preliminary qualitative analysis of the collected spectra was carried out by means of the LASSO method, after collecting the absorption spectra of the six kinds of one-component Fourier transform infrared (FTIRIRs) and a mixture of C _ 2H _ 4 and NH _ 3, combined with the self-built spectral database in the laboratory. Then the interference components are eliminated by LCLS method. The experimental results show that the method of LASSO combined with LCLS can effectively identify the target components in the spectrum, and most of the interference components can be eliminated even in the spectral bands with serious interference.
【作者單位】： 中國科學院安徽光學精密機械研究所 中國科學院環(huán)境光學與技術重點實驗室;中國科學技術大學;
【基金】：國家重點研發(fā)計劃項目(2016YFC0803001-08) 國家重大科學儀器設備開發(fā)專項(2013YQ22064302) 中國科學院前沿科學重點研究項目(QYZDY-SSW-DQC016) 國家自然科學基金項目(41405029)資助
【分類號】：O433.4
，

本文編號：2007775