天津农学院学报 ›› 2022, Vol. 29 ›› Issue (4): 60-64.doi: 10.19640/j.cnki.jtau.2022.04.012

• 研究与简报 • 上一篇    下一篇

掺杂婴幼儿奶粉的中红外光谱检测

黄明月, 赵晨, 吴海云, 靳皓, 杨仁杰通信作者   

  1. 天津农学院 工程技术学院,天津 300392
  • 收稿日期:2021-06-28 出版日期:2022-12-30 发布日期:2023-01-03
  • 通讯作者: 杨仁杰(1978-),男,教授,博士,主要从事光谱检测技术与应用方面的研究。E-mail:rjyang1978@163.com。
  • 作者简介:黄明月(1998-),女,硕士在读,主要从事光谱检测方面的研究。E-mail:937851273@qq.com。
  • 基金资助:
    天津市自然科学基金项目(18JCYBJC96400,16JCQNJC08200); 国家自然科学基金项目(41771357,21607114,81471698)

Detection of adulterated infant milk powder by mid-infrared spectroscopy

Huang Mingyue, Zhao Chen, Wu Haiyun, Jin Hao, Yang RenjieCorresponding Author   

  1. College of Engineering and Technology, Tianjin Agricultural University, Tianjin 300392, China
  • Received:2021-06-28 Online:2022-12-30 Published:2023-01-03

摘要: 基于中红外衰减全反射光谱建立了掺杂婴幼儿奶粉定性定量分析模型。准备40个纯奶粉样品,并制备了同时掺杂奶粉样品(质量浓度为0.1%~30%淀粉和0.1%~40%面粉),采集所有样品的中红外衰减全反射光谱。将中红外光谱与偏最小二乘法结合建立掺杂奶粉的定性判别模型,对未知样品的判别正确率为94%。最后,选择波数962~1 122 cm-1建模区间,建立奶粉中掺杂淀粉的定量分析模型;选择波数3 522~3 362 cm-1,1 282~800 cm-1为建模区间,建立奶粉中掺杂面粉的定量分析模型。对于淀粉,模型对未知样品预测相关系数R为0.98,预测均方根误差为0.062;对于面粉,模型对未知样品预测相关系数R为0.99,预测均方根误差为0.032。

关键词: 婴幼儿奶粉, 中红外光谱, 淀粉, 面粉, 定性定量分析

Abstract: A qualitative and quantitative analysis model for adulterated infant milk powder was established based on mid-infrared attenuated total reflectance(ATR)spectroscopy. Forty pure milk powder samples were prepared, and adulterated milk powder samples(the mass concentrations were 0.1%-30% for starch and 0.1%-40% for flour)were prepared. The mid-infrared ATR spectra of all samples were collected. The identification model of adulterated milk powder was established by combining mid-infrared spectroscopy and partial least squares, and the accuracy of identification for unknown samples was 94%. Finally, the spectra in the range of 962-1 122 cm-1 was selected to establish the quantitative analysis model of starch in adulterated milk powder. The spectra in the range of 3 522-3 362 cm-1 and 1 282-800 cm-1 were selected to establish the quantitative analysis model of flour in adulterated milk powder. For starch, the correlation coefficient R of prediction model for the unknown sample was 0.98, and the root mean square error of prediction(RMSEP)was 0.062. For flour: the correlation coefficient R of prediction model for the unknown sample was 0.99, and the RMSEP was 0.032.

Key words: infant milk powder, mid-infrared spectrum, starch, flour, qualitative and quantitative analysis

中图分类号: