首页 | 本学科首页   官方微博 | 高级检索  
     

拉曼光谱定量分析乙醇含量的非线性回归方法研究
引用本文:孙兰君,张延超,任秀云,付石友,田兆硕. 拉曼光谱定量分析乙醇含量的非线性回归方法研究[J]. 光谱学与光谱分析, 2016, 36(6): 1771-1774. DOI: 10.3964/j.issn.1000-0593(2016)06-1771-04
作者姓名:孙兰君  张延超  任秀云  付石友  田兆硕
作者单位:哈尔滨工业大学(威海)信息与光电子研究所,山东 威海 264209
基金项目:国家自然科学基金项目(41306092),山东省自然科学基金项目(ZR2013DQ026),山东省科技发展计划项目(2011GHY11514),中央高校基本科研业务费专项资金项目(HIT .NSRIF .2016087)
摘    要:传统拉曼特征峰峰比法一般采用线性回归法建立乙醇浓度与峰峰比的线性关系从而反演乙醇浓度实现乙醇定量分析,但仅在较低浓度范围适用。针对这一问题,采用自主研制的激光拉曼乙醇含量检测系统实验研究了不同浓度乙醇溶液拉曼光谱特征峰(非对称CH2伸缩振动2 924.0 cm-1)与本底水峰(3 350 cm-1)相对强度关系,提出适用于大范围乙醇浓度测量的非线性回归分析方法。利用邻域平均算法去除拉曼光谱突变噪声,结合多点插值处理实现光谱基线校准。基线校准及归一化处理后,可有效消除突变噪声及强荧光背景的影响。分别采用二次多项式和e指数数学模型对拉曼峰值强度比随乙醇浓度变化关系进行非线性回归并与线性回归分析进行对比。结果表明,线性拟合相关系数约为0.991,线性回归模型乙醇浓度准确测量的适用范围为15%~60%;非线性拟合相关系数高于0.997,非线性回归模型乙醇浓度精确测量的适用范围为3%~97%。非线性数学模型可为乙醇溶液浓度定量分析提供理论基础,将该数学模型应用于乙醇含量检测系统,可实时反演较为精确的乙醇浓度,从而实现大浓度范围内具有荧光背景干扰的乙醇溶液快速、实时、准确的定量分析。

关 键 词:激光拉曼光谱  乙醇浓度  非线性回归  定量分析  
收稿时间:2015-02-03

Nonlinear Regression Methods of Ethanol Raman Spectra Quantitative Analysis
SUN Lan-jun,ZHANG Yan-chao,REN Xiu-yun,FU Shi-you,TIAN Zhao-shuo. Nonlinear Regression Methods of Ethanol Raman Spectra Quantitative Analysis[J]. Spectroscopy and Spectral Analysis, 2016, 36(6): 1771-1774. DOI: 10.3964/j.issn.1000-0593(2016)06-1771-04
Authors:SUN Lan-jun  ZHANG Yan-chao  REN Xiu-yun  FU Shi-you  TIAN Zhao-shuo
Affiliation:Information Optoelectronics Research Institute, Harbin Institute of Technology, Weihai, Weihai 264209, China
Abstract:Ethanol concentration quantitative analysis of ethanol‐water solution can be realized by measuring the ratio of Raman characteristic peak heights .The content of ethanol can be determined by linear relation between relative intensity ratio and etha ‐nol concentration .However ,this analytical method only applies to the ethanol solution at low concentration .Concerning this is‐sue ,relative intensity of characteristic peak of ethanol (asymmetric stretching vibration of CH 2 2 924.0 cm - 1 ) and peak back‐ground of water (3 350 cm - 1 ) at different ethanol concentration is experimentally measured by using a self‐developed laser Raman ethanol content detection system .According to the relationship between relative ratio of characteristic peak heights and ethanol concentration ,the nonlinear regression analysis methods are proposed to apply in the measurement of ethanol concentra ‐tion in a wide range .Adjacent region average method is utilized to remove mutational random noise in Raman spectra of ethanol solution .Combined with multi‐point interpolation processing ,the baseline of Raman spectra can be calibrated .The influences of mutational random noise and the strong fluorescence background can be effectively eliminated with baseline correction and nor ‐malization methods .Polynomial and exponential mathematical models are adopted for nonlinear regression analyses by the rela ‐tion between ratio of characteristic peak heights and concentration of ethanol solution .The analysis results show that the correla‐tion coefficient of linear fitting and nonlinear fitting is about 0.991 and higher than 0.997 respectively .The linear analytical method can be effectively applied when ethanol concentration range is 15% ~ 60% .The nonlinear analytical method has higher measurement accuracy in a wider ethanol concentration range of 3% ~ 97% .Nonlinear mathematical model will provide theoreti‐cal basis for analysis of ethanol concentration ,which can be applied in laser Raman ethanol content detection system to calculate the relatively accurate ethanol concentration of ethanol‐water solution .Rapid ,real‐time and accurate quantitative analysis of wide concentration range ethanol solution ,which has mutational random noise and strong fluorescence background interference ,can be achieved by these analytical methods .
Keywords:Laser Raman spectra  Ethanol  Nonlinear regression methods  Quantitative analysis
本文献已被 CNKI 万方数据 等数据库收录!
点击此处可从《光谱学与光谱分析》浏览原始摘要信息
点击此处可从《光谱学与光谱分析》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号