摘要
提出了一种基于拉曼光谱技术(Raman)和激光诱导击穿光谱技术(LIBS)的多光谱特征融合技术(MFFT),利用拉曼光谱中分子组分信息和激光诱导击穿光谱中原子组分信息之间的互补特性,采用自适应小波变换(AWT)-竞争性自适应加权(CARS)-偏最小二乘回归(PLS)建模技术,获取了面粉体系更为全面的特征信息。在多光谱特征融合技术中,首先采用AWT-CARS方法分别提取拉曼光谱和激光诱导击穿光谱中的特征变量,然后将两者的特征变量融合为一个向量,采用PLS方法构建MFFT模型,实现了面粉掺杂物的定量分析。通过对二氧化钛、硫酸铝钾等面粉掺杂体系建模分析,考察MFFT模型的有效性。结果表明,与单一拉曼光谱技术或激光诱导击穿光谱技术建立的预测模型相比,MFFT模型显著提升了模型的预测性能,二氧化钛和硫酸铝钾预测模型的线性相关系数分别从相对较差的Raman模型的0.884、0.877提升到0.981、0.980,其预测均方根误差分别从相对较差的Raman模型的0.151、0.154降低到0.069、0.068。表明多光谱特征融合技术可以准确提取Raman光谱中的分子信息和LIBS光谱中的元素信息,使其互为补充、互为校正,进而有效克服面粉基质对掺杂组分定量分析的干扰,显著提高模型的预测精度。
A multispectral features fusion technique (MFFT ) was developed for the quantitative analysis of adulterated components in flour samples in this paper.The MFFT consisted of iterative wavelet transform (AWT ),competitive adaptive reweighted sampling (CARS ) and partial least squares regression (PLS ),utilizing the molecular information in Raman spectroscopy and the atomic information in laser-induced breakdown spectroscopy (LIBS ) to extract important features efficiently.In the MFFT,AWT and CARS were combined to select features in the Raman spectra and LIBS spectra,respectively,which were fused for PLS modeling.As a result,the MFFT model was generated to quantitate the adulterant contents in flour samples.To verify the performance of the MFFT method,the titanium dioxide and aluminum potassium sulfate in flour samples were quantitated.When compared with the prediction models based on individual spectroscopy,the correction coefficients of the MFFT models for titanium dioxide and aluminum potassium sulfate were significantly improved from 0.884,0.877 of Raman models to 0.981,0.980,whose root mean square errors were also decreased from 0.151,0.154 of Raman models to 0.069,0.068,respectively.Results indicated that the MFFT is a promising tool for quantitative analysis of adulterants in flour samples,extracting the molecular information in Raman spectra and the element information in LIBS spectrua accurately.The MFFT could make Raman spectroscopy and LIBS spectroscopy complementary for mutual correction,suppressing the effects of flour matrix on the quantitative analysis of adulterated components effectively,improving the prediction accuracy of the model significantly.
作者
刘丰奎
张翠
黄志轩
刘盼西
陈达
LIU Feng-kui;ZHANG Cui;HUANG Zhi-xuan;LIU Pan-xi;CHEN Da(Shanghai AnJie Environmental Protection Science & Technology Co.,LTD.,Shanghai 201906,China;School of Precision Instrument and Opto-Electronics Engineering,Tianjin University,Tianjin 300072,China)
出处
《分析测试学报》
CAS
CSCD
北大核心
2019年第4期390-395,共6页
Journal of Instrumental Analysis
基金
国家重点研发计划(2018YFF01011703)
国家自然科学基金(61378048
21305101)
关键词
拉曼光谱
激光诱导击穿光谱
多光谱特征融合技术
面粉掺杂
Raman spectroscopy
laser-induced breakdown spectroscopy
multispectral feature fusion technology
flour adulteration
