掺伪珍珠粉的近红外光谱定量鉴别方法  

Quantitative identification of adulterated pearl powder using near- infrared spectroscopy

作　　者：张轩[1] 白桦[1] 李文涛[1] 阮华君 廖杰 闫妍[1] 席广成[1] 

机构地区：[1]中国检验检疫科学研究院工业与消费品安全研究所,北京100123 [2]浙江长生鸟珍珠生物科技有限公司,浙江诸暨311804

出　　处：《现代化工》2014年第11期155-157,159,共4页Modern Chemical Industry

基　　金：质检公益性科研专项(2012104013-10)

摘　　要：采用偏最小二乘法建立掺杂贝壳粉的珍珠粉的近红外光谱定量分析模型.以掺杂不同比例贝壳粉和珍珠粉作为掺伪样本,采集样本在4 000～10 000 cm-1范围内的近红外漫反射光谱,采用交叉验证和外部验证考察所建模型的可靠性.优化建模参数为：波数范围为4 100～7 500 cm-1;光谱预处理方法为二阶导数/矢量归一化（SNV）/NorrisDerivative（5,5）滤波.所建定量分析模型的校正相关系数为0.9997,校正均方差为0.708,交叉验证均方差为3.15.实验证明近红外光谱技术可用于掺伪珍珠粉的定量快速鉴别.Near-infrared spectroscopy quantitative analysis model for pearl powder adulterated with shell powder is established by partial least squares. The near-infrared diffuse reflectance spectra of mixed samples of pearl powder and shell powder with different proportions are collected in 4 000 - 10 000 cm-1 region. The reliability of the established model is verified by cross-validation and external test. The results show that the optimum wavenumber range is 4 I00 -7 500 cm-1 and the optimum spectra pre-treatment method is second derivative/standard normal variate （SNV）/Norris derivative filter （5,5）. The calibrate correlation coefficient of the quantitative analysis model is 0. 999 7, and the root mean square error of calibration is 0. 708 ,while the root mean square error of cross-validation is 3.15. The experiments prove that near-infrared spectroscopy technique can be used for rapid quantitative identification of adulterated pearl powder.

关 键 词：珍珠粉 贝壳粉 近红外光谱 偏最小二乘法 

分 类 号：R284.1[医药卫生—中药学]