基于遗传算法与线性叠加模型的混合物组成拉曼光谱定量分析

引用本文: 温国基, 戴连奎, 刘薇. 基于遗传算法与线性叠加模型的混合物组成拉曼光谱定量分析[J]. 分析化学, 2021, 49(1): 85-94. doi: 10.19756/j.issn.0253-3820.201500 shu

Citation: WEN Guo-Ji, DAI Lian-Kui, LIU Wei. Raman Spectroscopic Quantitative Analysis Based on Genetic Algorithm and Linear Superposition Principle[J]. Chinese Journal of Analytical Chemistry, 2021, 49(1): 85-94. doi: 10.19756/j.issn.0253-3820.201500 shu

基于遗传算法与线性叠加模型的混合物组成拉曼光谱定量分析

浙江大学工业控制技术国家重点实验室, 杭州 310027

通讯作者: 戴连奎,E-mail:lkdai@zju.edu.cn

基金项目:
国家自然科学基金项目（No.U1609213）资助。

摘要: 提出了一种基于遗传算法（Genetic algorithm，GA）与线性叠加模型的拉曼光谱定量分析算法。仅需已知一种主要物质的拉曼光谱，即可基于一批包括该物质的混合物样本，结合相关分析获得各组分的特征峰位；再利用遗传算法拟合优化混合物对应的纯物质光谱矩阵；最后利用拟合得到的光谱矩阵，建立混合物中各种物质的拉曼光谱定量分析模型。本方法只在模型训练过程优化模型参数时计算量较大，而在实际预测时只需进行简单的矩阵运算即可，实时计算量很小，适用于连续在线应用。本方法可应用于液相或气相混合物的定量分析。以天然气为例，仅知主要组分甲烷的拉曼光谱时，基于一组已知其组成与拉曼光谱的天然气训练样本，建立了天然气中主要组分（CH₄、C₂H₆、C₃H₈、CO₂、N₂、H₂、CO）拉曼光谱定量分析模型；对于测试样本集，对应的预测均方根误差（Root mean square error，RMSE）分别为0.488%、0.103%、0.358%、0.079%、0.087%、0.126%和0.022%，复相关系数（Multi-correlation coefficient，R²）分别为0.995、0.991、0.990、0.991、0.998、0.999和0.999。尽管此结果不及所有纯物质光谱均已知情况下得到的结果，但为大多数纯物质光谱未知的情况提供了一种解决方案。

关键词:

English

Raman Spectroscopic Quantitative Analysis Based on Genetic Algorithm and Linear Superposition Principle

College of Control Science and Engineering, Zhejiang University, Hangzhou 310027, China

Corresponding author: DAI Lian-Kui, lkdai@zju.edu.cn

Received Date: 16 August 2020
Revised Date: 04 November 2020
Fund Project:
Supported by the National Natural Science Foundation of China (No.U1609213).

Abstract: A Raman quantitative analysis algorithm based on genetic algorithm (GA) and linear additive model was developed. When the Raman spectrum of one main substance in a batch of mixed samples was known, the characteristic peak position of each component could be obtained by correlation analysis. Then genetic algorithm (GA) was used to fit and optimize the corresponding spectral matrix of the pure substance. Finally, the obtained spectral matrix was used to establish the Raman quantitative analysis model of various substances in the mixture. This method only required a large amount of calculation when optimizing model parameters in the model training process, and only needed to perform simple matrix operations in actual prediction, showing a small real-time calculation amount. The developed method was suitable for continuous online applications, and could be applied to the quantitative analysis of liquid and gas mixture. By taking natural gas as an example, when the Raman spectrum of the main component in methane was known, based on a set of natural gas training samples whose composition and Raman spectrum were known, a quantitative analysis model based on Raman spectrum for main components in natural gas (CH₄, C₂H₆, C₃H₈, CO₂, N₂, H₂, CO) was developed. For the test sample set, the corresponding root mean square error (RMSE) were 0.488%, 0.103%, 0.358%, 0.079%, 0.087%, 0.126% and 0.022%, and the multiple correlation coefficient (R²) were 0.995, 0.991, 0.990, 0.991, 0.998, 0.999 and 0.999, respectively. Although the result was just slightly inferior to the result when the spectra of all pure substances were known, the algorithm presented here provided a solution for the case when some pure substance spectra were unknown.

Key words:

1. [1]
  ESTEKI M, SHAHSAVARI Z, SIMAL-GANADARA J. Food Res. Int., 2019, 122:303-317.
2. [2]
  JOHANNES K. Energies, 2015, 8(4):3165-3197.
3. [3]
  KIEFER J, SEEGER T, STEUER S, SCHORSCH S, WEIKL M C, LEIPERTZ A. Meas. Sci. Technol., 2008, 19(8):1-9.
4. [4]
  WU Lin, OUYANG Zhao-Hui, CAO Shu-Chao, YI De-Lian, SUN Shao-Xue, LIU Xia. J. Light Scattering, 2005, 2:180-186伍林, 欧阳兆辉, 曹淑超, 易德莲, 孙少学, 刘峡. 光散射学报, 2005, 2:180-186.
5. [5]
  MARTEAU P, ZANIERSZYDLOWSKI N, AOUFI A, HOTIER G, CANSELL F. Vib. Spectrosc., 1995, 9(1):101-109.
6. [6]
  SHARMA R, POONACHA S, BEKAL A, VARTAK S, WELING A, TILAK V, MITRA C. Opt. Eng., 2016, 55(10):104103.
7. [7]
  NUMATA Y, SHINOHARA Y, TANAKA H. Spectrosc. Lett., 2019, 52(5):306-311.
8. [8]
  GAO Ying, DAI Lian-Kui, ZHU Hua-Dong, CHEN Yun-Liang, ZHOU Li. Chin. J. Anal. Chem., 2019, 47(1):67-76. 高颖, 戴连奎, 朱华东, 陈昀亮, 周理. 分析化学, 2019, 47(1):67-76.
9. [9]
  CHEN Y L, DAI L K. Anal. Sci., 2019, 35(5):511-515.
10. [10]
  AIDILA J A, SOARES F L F, FARIAS M A D, CARNEIRO R L. Anal. Lett., 2017, 50(7):1126-1138.
11. [11]
  BONOLDI L, FRIGERIO F, DI PAOLO L, SAVOIN A, BARBIERI D, GRIGO D. Energy Fuels, 2018, 32(9):8955-8965.
12. [12]
  KARUNATHILAKA S R, YAKES B J, FARRIS S, MICHAEL T J, HE K Q, CHUNG J K, SHAH R, MOSSOBA M M. Food Anal. Methods, 2018, 11(4):969-979.
13. [13]
  LIU Wei, DAI Lian-Kui. Chin. J. Anal. Chem., 2019, 47(1):77-85. 刘薇, 戴连奎. 分析化学, 2019, 47(1):77-85.
14. [14]
  HE S X, FANG S X, LIU X H, ZHANG W, XIE W Y, ZHANG H, WEI D P, FU W L, PEI D S. Chemom. Intell. Lab. Syst., 2016, 15:1-9.
15. [15]
  LI Z, GUAN A H, GE H Y, LIAN F Y. Microchem. J., 2017, 132:185-189.
16. [16]
  GE Ji-Ke, QIU Yu-Hui, WU Chun-Ming, PU Guo-Lin. Appl. Res. Comput., 2008, 10:2911-2916. 葛继科, 邱玉辉, 吴春明, 蒲国林. 计算机应用研究, 2008, 10:2911-2916.
17. [17]
  ZHANG Guo-Ming. Sci. Technol. Vision, 2013, 09:37+36. 张国民. 科技视界, 2013, 09:37+36.
18. [18]
  KOSEK F, CULKA A, ROUSAKI A, VANDENABEELE P, JEHLICKA J. Spectrochim. Acta A, 2020, 243:118818.
19. [19]
  KUYLIEV B T, JUMABOEV A J, KHUSHVAKTOV K A, MEYLIEV L O, ABOSONOV A A, HUDOYBERDIEV B G. Ukrainian J. Phy., 2019, 64(1):72-78
20. [20]
  KE Yi-Kan, DONG Hui-Ru. Handbook of Analytical Chemistry 3B Molecular Spectroscopy Analysis, Chemical Industry Press, 2016:898-899. 柯以侃, 董慧茹. 分析化学手册3B分子光谱分析(第三版), 化学工业出版社, 2016:898-899.

扫一扫看文章

计量

PDF下载量: 4
文章访问数: 28
HTML全文浏览量: 8

通讯作者: 陈斌, bchen63@163.com

1.
沈阳化工大学材料科学与工程学院沈阳 110142

基于遗传算法与线性叠加模型的混合物组成拉曼光谱定量分析

通讯作者: 戴连奎,E-mail:lkdai@zju.edu.cn

English

Raman Spectroscopic Quantitative Analysis Based on Genetic Algorithm and Linear Superposition Principle

Corresponding author: DAI Lian-Kui, lkdai@zju.edu.cn

CCS Chemistry：嵌段共聚物自组装成 “三明治”二维有机材料，实现纳米级压力传感功能

CCS Chemistry：颜徐州、鲍哲南：你的皮肤可能是一片SPMs

CCS Chemistry：工作高效不疲劳，只须让催化剂动起来

CCS Chemistry：静电组装导向聚合- 聚电解质纳米凝胶量化制备新策略

CCS Chemistry：金黄色葡萄球菌醛基脱氢酶是如何识别特异性底物的？

计量

通讯作者: 陈斌, bchen63@163.com

中国化学会秘书处

基于遗传算法与线性叠加模型的混合物组成拉曼光谱定量分析

通讯作者: 戴连奎,E-mail:lkdai@zju.edu.cn

English

Raman Spectroscopic Quantitative Analysis Based on Genetic Algorithm and Linear Superposition Principle

Corresponding author: DAI Lian-Kui, lkdai@zju.edu.cn

CCS Chemistry：嵌段共聚物自组装成 “三明治”二维有机材料，实现纳米级压力传感功能

CCS Chemistry：颜徐州、鲍哲南： 你的皮肤可能是一片SPMs

CCS Chemistry：工作高效不疲劳，只须让催化剂动起来

CCS Chemistry：静电组装导向聚合- 聚电解质纳米凝胶量化制备新策略

CCS Chemistry：金黄色葡萄球菌醛基脱氢酶是如何识别特异性底物的？

计量

文章相关

通讯作者: 陈斌, bchen63@163.com

中国化学会秘书处

CCS Chemistry：颜徐州、鲍哲南：你的皮肤可能是一片SPMs