多级同时成分分析用于温控近红外光谱数据的处理
详细信息 查看官网全文
- 英文论文题名：Multilevel Simultaneous Component Analysis of Temperature Dependent Near Infrared Spectra
- 论文作者：崔晓宇 ; 蔡文生 ; 邵学广
- 英文论文作者：Xiaoyu Cui ; Wensheng Cai ; Xueguang Shao ; Research Center for Analytical Sciences ; College of Chemistry ; Nankai University
- 年：2016
- 作者机构：南开大学化学学院分析科学研究中心;
- 论文关键词：温控近红外光谱 ; 多级同时成分分析 ; QSTR模型 ; 定量分析
- 会议召开时间：2016-07-01
- 会议录名称：中国化学会第30届学术年会摘要集-第二十五分会：化学信息学与化学计量学
- 语种：中文
- 分类号：O657.33
- 学会代码：ZGHY
- 会议名称：中国化学会第30届学术年会-第二十五分会 ; 化学信息学与化学计量学
- 会议地点：中国辽宁大连
- 主办单位：中国化学会
- 学会名称：中国化学会
- 页数：1
- 文件大小：138k
- 原文格式：D
- 会议级别：全国
摘要
近红外光谱容易受到温度等实验条件的影响,利用这种影响,本课题组建立了温度-光谱定量分析(QSTR)模型,并以此为基础,预测了溶液的组成~([1,2])。为了进一步提取温度信息,使用多级同时成分分析(MSCA)算法对温控近红外光谱数据进行了解析~([3])。MSCA方法能够在不同层次分析数据,并且分别考察数据内存在的不同类型的方差。利用该算法可增强模型解释性的特点,分析了混合溶液的温控近红外光谱数据,分别建立了温度间模型和温度内模型。采用温度间模型得到的温度系数建立了QSTR模型;采用温度内模型得到的浓度系数预测了溶剂的组成。为提高方法的适用性,本文对该方法进行了优化。使用MSCA算法分析了35~45℃范围内,浓度范围为0.2~2.0 g L~(-1)的葡萄糖水溶液样品的近红外光谱数据,结果表明MSCA算法适用于人体温度范围内低浓度葡萄糖含量的预测。
Near-infrared(NIR) spectra are sensitive to the variation of experimental conditions,such as temperature.In our previous works,the relationship between NIR absorption spectra and temperature was quantitatively analyzed(QSTR) and applied to the quantitative determination of the compositions in mixtures.For further extracting information,multilevel simultaneous component analysis(MSCA) was used to investigate temperature dependent NIR spectra.MSCA has an advantage of summarizing the different types of variation at different levels in the data.Therefore,a between-temperature model and a within-temperature model was built.QSTR was established by the temperature coefficients in the former model,and quantitative analysis was achieved by concentration coefficients in the latter model.In this work,MSCA was further studied using the spectra of aqueous glucose samples in the temperature range of 35~45 ℃ and the concentration range of 0.2~2.0 g L~(-1).The results show that the quantitative analysis can be well achieved.
Near-infrared(NIR) spectra are sensitive to the variation of experimental conditions,such as temperature.In our previous works,the relationship between NIR absorption spectra and temperature was quantitatively analyzed(QSTR) and applied to the quantitative determination of the compositions in mixtures.For further extracting information,multilevel simultaneous component analysis(MSCA) was used to investigate temperature dependent NIR spectra.MSCA has an advantage of summarizing the different types of variation at different levels in the data.Therefore,a between-temperature model and a within-temperature model was built.QSTR was established by the temperature coefficients in the former model,and quantitative analysis was achieved by concentration coefficients in the latter model.In this work,MSCA was further studied using the spectra of aqueous glucose samples in the temperature range of 35~45 ℃ and the concentration range of 0.2~2.0 g L~(-1).The results show that the quantitative analysis can be well achieved.
引文
[1]Shao,X.G.;Kang,J.;Cai,W.S.Talanta.2010,82(3):1017.
[2]Kang,J.;Shao,X.G.;Cai,W.S.Talanta.2011,85(3):420.
[3]Shan,R.F.;Zhao,Y.;Fan,M.L.Liu,X.W.;Cai,W.S.;Shao,X.G.Talanta,2015,131(131):170.
[2]Kang,J.;Shao,X.G.;Cai,W.S.Talanta.2011,85(3):420.
[3]Shan,R.F.;Zhao,Y.;Fan,M.L.Liu,X.W.;Cai,W.S.;Shao,X.G.Talanta,2015,131(131):170.