• Spectroscopy and Spectral Analysis
  • Vol. 42, Issue 1, 164 (2022)
Rong-ke YE*, Qing-chen KONG1;, Dao-liang LI1; 2;, Ying-yi CHEN1; 2;, Yu-quan ZHANG1;, and Chun-hong LIU1; 2; *;
Author Affiliations
  • 1. College of Information and Electrical Engineering, China Agricultural University, Beijing 100083, China
  • show less
    DOI: 10.3964/j.issn.1000-0593(2022)01-0164-06 Cite this Article
    Rong-ke YE, Qing-chen KONG, Dao-liang LI, Ying-yi CHEN, Yu-quan ZHANG, Chun-hong LIU. Shrimp Freshness Detection Method Based on Broad Learning System[J]. Spectroscopy and Spectral Analysis, 2022, 42(1): 164 Copy Citation Text show less
    Architecture of the broad learning system
    Fig. 1. Architecture of the broad learning system
    Average spectral curves with standard deviation
    Fig. 2. Average spectral curves with standard deviation
    Visualization analysis(a), (b), (c), (d): Spectral curves by using different preprocessing methods; (e), (f), (g), (h): Visualization using t-SNE
    Fig. 3. Visualization analysis
    (a), (b), (c), (d): Spectral curves by using different preprocessing methods; (e), (f), (g), (h): Visualization using t-SNE
    Analysis of the importance of wavelength variables by RF
    Fig. 4. Analysis of the importance of wavelength variables by RF
    Analysis of effective PC scores and loading(a): PCA score plot of PC1vs. PC2; (b): Wavelength selection on PC1 and PC2 loading lines
    Fig. 5. Analysis of effective PC scores and loading
    (a): PCA score plot of PC1vs. PC2; (b): Wavelength selection on PC1 and PC2 loading lines
    The 2D-COS spectrum of samples with different days of refrigeration(a): Synchronous contourmap plot; (b): Autocorrelation peak intensity curve
    Fig. 6. The 2D-COS spectrum of samples with different days of refrigeration
    (a): Synchronous contourmap plot; (b): Autocorrelation peak intensity curve
    ModelParameterAccuracy of calibration setAccuracy of prediction set
    RF-PLS-DA1391.75%(289/315)90.48%(95/105)
    RF-ELM3495.56%(301/315)95.24%(100/105)
    RF-BLS(2-30, 0.9)98.41%(310/315)97.14%(102/105)
    PCA-PLS-DA590.48%(285/315)89.52%(94/105)
    PCA-ELM2193.97%(296/315)91.43%(96/105)
    PCA-BLS(2-20, 0.9)96.19%(303/315)94.29%(99/105)
    2D-COS-PLS-DA285.10%(269/315)81.90%(86/105)
    2D-COS-ELM1489.21%(281/315)87.62%(92/105)
    2D-COS-BLS(2-20, 0.7)91.43%(288/315)89.52%(94/105)
    Table 1. Analysis of modeling results based on feature wavelengths
    Rong-ke YE, Qing-chen KONG, Dao-liang LI, Ying-yi CHEN, Yu-quan ZHANG, Chun-hong LIU. Shrimp Freshness Detection Method Based on Broad Learning System[J]. Spectroscopy and Spectral Analysis, 2022, 42(1): 164
    Download Citation