Spectroscopy and Spectral Analysis, Volume. 41, Issue 11, 3377(2021)

Prediction Model of TVB-N Concentration in Mutton Based on Near Infrared Characteristic Spectra

Xu ZHANG1、1;, Xue-bing BAI1、1;, Xue-pei WANG2、2;, Xin-wu LI2、2;, Zhi-gang LI3、3;, and Xiao-shuan ZHANG2、2; 4; *;
Author Affiliations
  • 11. College of Information and Electrical Engineering, China Agricultural University, Beijing 100083, China
  • 22. College of Engineering, China Agricultural University, Beijing 100083, China
  • 33. College of Information Science and Technology, Shihezi University, Shihezi 832003, China
  • show less
    Figures & Tables(16)
    The NIR spectra of mutton samples
    The outlier results detected by Monte Carlo sampling method
    Distribution of Mahalanobis distance of mutton samples
    Characteristic wavelength selection process using CARS
    Ridge trace of ridge regression analysis
    Characteristic wavelength selection using IUVE
    Wavelength selection based on SPA
    Characteristic wavelength selection based on SPA
    Prediction results of TVB-N concentration in mutton by CARS-PLS model
    Prediction results of TVB-N concentration in mutton by CARS-SVM model
    Prediction results of TVB-N concentration in mutton by CARS-LS-SVM model
    • Table 1. Statistical results of TVB-N concentration of mutton samples

      View table
      View in Article

      Table 1. Statistical results of TVB-N concentration of mutton samples

      样本集样本数TVB-N含量/(mg·100 g-1)
      最大值最小值平均值标准差
      校正集8726.584.8114.515.82
      验证集2923.265.5512.665.07
    • Table 2. Comparison of PLS prediction models with different pretreatment methods

      View table
      View in Article

      Table 2. Comparison of PLS prediction models with different pretreatment methods

      预处理方法主成
      校正集验证集
      RC2RMSECRV2RMSEV
      RAW110.848 12.253 40.808 52.180 9
      SNV90.710 32.509 70.654 63.207 9
      MSC120.775 52.293 80.696 63.296 2
      Normalization100.839 32.317 70.798 52.237 3
      Centering90.840 62.308 60.814 82.144 8
      Autoscaling80.705 12.515 20.694 03.290 2
      MAS120.839 92.313 70.814 62.146 0
      SGS110.850 22.211 50.824 62.105 8
    • Table 3. Comparison of PLS models with different wavelength selection methods

      View table
      View in Article

      Table 3. Comparison of PLS models with different wavelength selection methods

      特征波长
      提取方法
      变量
      个数
      主成
      校正集验证集
      RC2RMSECRV2RMSEV
      Full spectrum1 92190.840 22.311 50.814 62.145 8
      CARS1490.890 01.917 80.873 61.771 6
      UVE70390.834 42.353 00.816 72.133 5
      IUVE14490.844 82.278 00.823 72.092 3
      SPA15100.826 12.411 40.798 42.237 7
    • Table 4. Comparison of SVM models with different wavelength selection methods

      View table
      View in Article

      Table 4. Comparison of SVM models with different wavelength selection methods

      特征波长
      提取方法
      参数
      (C)
      参数
      (γ)
      校正集验证集
      RC2RMSECRV2RMSEV
      Full spectrum16.000 00.003 90.926 81.565 00.826 12.078 2
      CARS48.502 91.000 00.939 11.426 70.801 62.219 7
      UVE9.189 60.006 80.900 51.824 20.823 32.095 2
      IUVE1.741 10.011 80.916 11.674 90.824 12.090 1
      SPA256.000 00.189 50.881 31.992 10.791 92.273 7
    • Table 5. Comparison of LS-SVM models with different wavelength selection methods

      View table
      View in Article

      Table 5. Comparison of LS-SVM models with different wavelength selection methods

      特征波长
      提取方法
      参数
      (γ)
      参数
      (σ2)
      校正集验证集
      RC2RMSECRV2RMSEV
      Full spectrum93.497 63 641.32 80.914 51.690 70.844 71.964 3
      CARS12.031 35.627 60.923 51.599 00.850 11.929 4
      UVE26.185 01 118.8100.901 31.817 00.855 11.896 9
      IUVE94.239 71.716.10.911 21.723 40.856 81.886 2
      SPA131.846 920.751 50.879 32.008 80.824 32.089 2
    Tools

    Get Citation

    Copy Citation Text

    Xu ZHANG, Xue-bing BAI, Xue-pei WANG, Xin-wu LI, Zhi-gang LI, Xiao-shuan ZHANG. Prediction Model of TVB-N Concentration in Mutton Based on Near Infrared Characteristic Spectra[J]. Spectroscopy and Spectral Analysis, 2021, 41(11): 3377

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Orginal Article

    Received: Sep. 1, 2020

    Accepted: --

    Published Online: Dec. 17, 2021

    The Author Email:

    DOI:10.3964/j.issn.1000-0593(2021)11-3377-08

    Topics