• Spectroscopy and Spectral Analysis
  • Vol. 41, Issue 12, 3688 (2021)
Shi-jie XIAO1、*, Qiao-hua WANG1、1; 2; *;, Yi-kai FAN3、3;, Rui LIU3、3;, Jian RUAN3、3;, Wan WEN4、4;, Ji-qi LI4、4;, Huai-feng SHAO4、4;, Wei-hua LIU5、5;, and Shu-jun ZHANG3、3; *;
Author Affiliations
  • 11. College of Engineering, Huazhong Agricultural University, Wuhan 430070, China
  • 33. Key Laboratory of Animal Breeding and Reproduction of Minstry of Education, Huazhong Agricultural University, Wuhan 430070, China
  • 44. Ningxia Hui Autonomous Region Animal Husbandary Workstation, Yinchuan 750002, China
  • 55. Ningxia Hui Autonomous Region Veterinary Medicine and Feed Suqervision Institute, Yinchuan 750011, China
  • show less
    DOI: 10.3964/j.issn.1000-0593(2021)12-3688-07 Cite this Article
    Shi-jie XIAO, Qiao-hua WANG, Yi-kai FAN, Rui LIU, Jian RUAN, Wan WEN, Ji-qi LI, Huai-feng SHAO, Wei-hua LIU, Shu-jun ZHANG. Rapid Determination of αs1-Casein and κ-Casein in Milk Based on Fourier Transform Infrared Spectroscopy[J]. Spectroscopy and Spectral Analysis, 2021, 41(12): 3688 Copy Citation Text show less
    Original spectra of cow’s milk
    Fig. 1. Original spectra of cow’s milk
    Average spectra of cow’s milk
    Fig. 2. Average spectra of cow’s milk
    Mean value and variance distributions of (a)αs1 and (b) κ-casein
    Fig. 3. Mean value and variance distributions of (a)αs1 and (b) κ-casein
    Variable selection of CARS
    Fig. 4. Variable selection of CARS
    Screening characteristic wavelengths by UVE
    Fig. 5. Screening characteristic wavelengths by UVE
    (a) RMSE; (b) Selected optimal wavelength number index
    Fig. 6. (a) RMSE; (b) Selected optimal wavelength number index
    Optimal model for αs1-casein based on CARS
    Fig. 7. Optimal model for αs1-casein based on CARS
    Optimal model for κ-casein based on UVE
    Fig. 8. Optimal model for κ-casein based on UVE
    指标αs1-酪蛋白/(g·L-1)κ-酪蛋白/(g·L-1)
    样本集训练集测试集训练集测试集
    最小值2.783 42.695 31.156 11.158 4
    最大值13.782 416.208 76.475 45.992 9
    平均值7.672 48.047 03.171 22.881 9
    标准差2.407 72.928 51.127 51.125 0
    Table 1. Data statistics of partitioning sample sets by SPXY algorithm
    波长选
    择算法
    预处理
    方式
    变量
    个数
    训练集测试集
    RcRMSECRpRMSEP
    CARS原始150.814 71.400 60.756 61.933 9
    SNV380.837 41.343 70.820 91.771 1
    MSC240.825 11.402 80.839 71.825 2
    一阶导数240.882 71.136 30.899 81.372 6
    二阶导数320.844 51.324 30.800 91.808 5
    normaliz180.801 91.447 90.794 31.825 6
    一阶差分110.833 61.341 00.817 81.746 2
    二阶差分150.877 71.150 80.767 91.883 3
    UVE原始690.825 71.370 10.838 81.644 1
    SNV1290.832 71.365 50.816 91.808 8
    MSC1370.812 41.420 00.800 61.887 5
    一阶导数1080.880 31.165 30.851 11.594 7
    二阶导数1000.852 61.310 50.845 21.701 5
    normaliz1010.840 71.327 90.819 11.746 1
    一阶差分950.866 41.246 60.835 31.711 8
    二阶差分1660.941 20.840 30.821 41.711 8
    SPA原始230.825 61.388 20.839 21.676 9
    SNV440.923 70.929 10.724 42.003 4
    MSC140.694 31.741 70.794 41.849 2
    一阶导数210.827 01.381 70.821 51.767 8
    二阶导数210.943 60.808 40.837 11.603 8
    normaliz400.777 61.532 10.743 11.974 3
    一阶差分90.888 21.129 30.827 01.694 1
    二阶差分60.798 81.465 30.831 01.707 0
    Table 2. SVR prediction model of αs1-casein based on 3 characteristic variable selection methods
    波长选
    择算法
    预处理
    方式
    变量
    个数
    训练集测试集
    RcRMSECRpRMSEP
    CARS原始330.883 60.529 50.792 50.685 4
    SNV150.824 50.649 40.793 60.692 4
    MSC200.845 50.605 60.826 90.651 3
    一阶导数150.872 40.552 80.837 20.614 1
    二阶导数150.885 20.524 40.761 40.726 0
    normaliz200.852 20.595 80.823 90.657 0
    一阶差分110.858 20.580 80.850 70.606 6
    二阶差分160.864 30.566 10.873 90.576 0
    UVE原始1450.842 70.610 10.828 90.637 0
    SNV1070.893 90.507 80.848 70.596 6
    MSC1140.881 00.538 80.848 60.605 4
    一阶导数570.860 30.591 00.851 40.618 9
    二阶导数670.881 40.541 50.857 90.591 3
    normaliz1200.860 80.576 20.843 60.612 8
    一阶差分550.880 80.534 50.890 30.535 4
    二阶差分1070.914 70.473 50.887 70.558 1
    SPA原始230.850 40.593 40.848 90.608 9
    SNV170.886 20.526 30.834 10.622 4
    MSC150.861 70.576 80.820 30.638 3
    一阶导数140.853 70.587 70.885 10.540 8
    二阶导数70.807 60.664 00.850 80.618 1
    normaliz90.825 90.639 10.859 10.608 1
    一阶差分90.892 10.513 00.808 20.668 1
    二阶差分70.807 60.664 00.850 80.618 1
    Table 3. SVR prediction model of κ-casein based on different characteristic variable selection methods
    Shi-jie XIAO, Qiao-hua WANG, Yi-kai FAN, Rui LIU, Jian RUAN, Wan WEN, Ji-qi LI, Huai-feng SHAO, Wei-hua LIU, Shu-jun ZHANG. Rapid Determination of αs1-Casein and κ-Casein in Milk Based on Fourier Transform Infrared Spectroscopy[J]. Spectroscopy and Spectral Analysis, 2021, 41(12): 3688
    Download Citation