CASIA OpenIR  > 毕业生  > 硕士学位论文
Alternative TitleResearch on Quantitative Analysis of Mixture Solution Infrared Spectrum
Thesis Advisor彭思龙
Degree Grantor中国科学院研究生院
Place of Conferral中国科学院自动化研究所
Degree Discipline计算机应用技术
Keyword红外光谱 多元非线性回归 非对称最小二乘 光谱差减 Infrared Spectrum Multiple Nonlinear Regression Asymmetric Least Squares (Asls) Spectral Subtraction
Abstract红外光谱分析技术具有快速、无损、成本低等优点,对固体、液体或气体样品,对单一组分的纯净物和多种组分的混合物都可以测定,已被广泛应用于石油化工、食品工业和制药工业等领域。混合溶液是红外光谱分析的常见对象,采用先分离再分析的方法需要消耗大量的时间,并且由于分子间的相互作用无法得到原始混合物完全真实的信息。混合溶液中分子间的相互作用使得吸光度和浓度之间不再是简单的线性关系,如何建立两者之间的非线性回归模型是定量分析的关键所在。另外,光谱差减作为一种常用的分离重叠谱带的方法,有助于提取目标光谱和突出特征峰,但当不同组分的光谱重叠非常严重时,传统的光谱差减方法不再适用,建立一种通用的光谱差减模型对于混合溶液分析具有重要意义。 论文的主要内容包括: 1、 研究了混合溶液中浓度和吸光度之间的关系,建立了以浓度为自变量,吸光度为因变量的非线性回归模型。首先,从简单的非线性模型入手,假设吸光度是关于浓度的二次多项式模型。鉴于主成分分析在光谱分析中的有效性,将主成分分析方法与非线性回归相结合,建立了先进行特征提取再进行非线性回归的多元校正模型。实验结果显示,与常用的线性多元校正模型——主成分回归和偏最小二乘回归相比较,非线性多元校正模型的预测能力更强。 2、 研究了光谱差减问题,提出了一种比较通用的光谱差减方法。根据红外光谱差减的特点和光谱固有的性质,建立了基于非对称最小二乘的光谱差减模型,与传统的光谱差减方法相比,该模型不依赖于参考峰,使得该模型在光谱差减中具有普适性。通过实验证明,对于传统方法适用的光谱差减,该方法在定性和定量分析上略优于传统方法。在传统方法不适用的情况,该模型仍然能够处理。鉴于多次光谱差减会降低差谱的可靠性,将被减组分单一的光谱差减模型推广到了被减组分多种的情况,建立了一种通用的光谱差减模型。实验结果显示,该模型能够较好的处理多种被减组分的情况,并在定量和定性分析上具有实际意义。
Other AbstractInfrared (IR) spectroscopy is widely used in industry and agriculture, for its advantages in rapid and undamaged analysis. It is applied to solid, liquid and gas samples, either pure or mixture. It is time consuming to analyze mixture by separation, without considering the interaction in mixture yet. Due to the interaction among molecular, the linear relationship between absorbance and concentration does not retain. Then, linear –based multivariate calibration techniques such as PLS regression and PCR are not optimal for mixture spectra. Therefore, modeling the nonlinear relationship is significant for the quantitative analysis of mixture solution. In addition, there is not a general method for spectral subtraction which is a fundamental approach to separate overlapping bands and extract target spectrum from mixture spectrum. Traditional methods can’t deal with the cases when scaling spectrum and target spectrum overlap seriously and multiple components are subtracted from sample spectra. In view of the limitations of traditional method, it is important to improve the spectral subtraction model. The content in this thesis includes the following aspects: In order to build nonlinear calibration model between absorbance and concentration, quadratic polynomial regression is proposed. Considering the advantage of feature extraction, PCA and quadratic polynomial regression are combined. The nonlinear calibration models are evaluated and compared to linear calibration models based on PCR and PLS regression. Experimental results demonstrate the prediction ability of nonlinear calibration model is a little better than that of PCR and PLS regression. A new spectral subtraction model based on asymmetric least squares (AsLS) is developed, which is according to the feature of spectral subtraction and nature of infrared spectrum. It is a global optimization algorithm independent of reference peak, so this procedure can deal with most of binary system (single component is subtracted), even when ideal reference peak can’t be found in scaling spectrum. In addition to binary system, multiple system is also common. Given scaling spectrum, the subtraction model of multiple system is the same as binary system. Then, the problem of multiple system comes into N-scaling spectrum problem. In binary system, the results of qualitative and quantitative analysis based on subtraction model are better than that of traditional method. In multiple system (multiple components are su...
Other Identifier200928014629089
Document Type学位论文
Recommended Citation
GB/T 7714
杨朝玲. 混合溶液红外光谱的定量分析研究[D]. 中国科学院自动化研究所. 中国科学院研究生院,2012.
Files in This Item:
File Name/Size DocType Version Access License
CASIA_20092801462908(2782KB) 暂不开放CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[杨朝玲]'s Articles
Baidu academic
Similar articles in Baidu academic
[杨朝玲]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[杨朝玲]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.