• Spectroscopy and Spectral Analysis
  • Vol. 41, Issue 1, 188 (2021)
Ying CHEN1、1, Yang-mei XU1、1, Yuan-jian DI1、1, Xing-ning CUI1、1, Jie ZHANG1、1, Xin-de ZHOU1、1, Chun-yan XIAO1、1, and Shao-hua LI1、1
Author Affiliations
  • 1[in Chinese]
  • 11. Hebei Province Key Laboratory of Test/Measurement Technology and Instrument, School of Electrical Engineering, Yanshan University, Qinhuangdao 066004, China
  • show less
    DOI: 10.3964/j.issn.1000-0593(2021)01-0188-06 Cite this Article
    Ying CHEN, Yang-mei XU, Yuan-jian DI, Xing-ning CUI, Jie ZHANG, Xin-de ZHOU, Chun-yan XIAO, Shao-hua LI. COD Concentration Prediction Model Based on Multi-Spectral Data Fusion and GANs Algorithm[J]. Spectroscopy and Spectral Analysis, 2021, 41(1): 188 Copy Citation Text show less
    COD water sample spectra(a): UV spectrum; (b): NIR spectrum
    Fig. 1. COD water sample spectra
    (a): UV spectrum; (b): NIR spectrum
    COD prediction model based on GANs algorithms(a): Data level fusion; (b): Feature level fusion
    Fig. 2. COD prediction model based on GANs algorithms
    (a): Data level fusion; (b): Feature level fusion
    Comparison of evaluation parameters of different quantitative prediction models(a): Correction sets; (b): Verification sets
    Fig. 3. Comparison of evaluation parameters of different quantitative prediction models
    (a): Correction sets; (b): Verification sets
    融合方法预处理特征提取筛选波段/nm校正集验证集
    R2RMSECV偏差R2RMSEP偏差
    LLDFDer-1S-G--0.9782.3560.3180.9153.9280.764
    MLDFDer-1S-GBiPLS820~952
    1 719~1 836
    0.9841.6590.0610.9332.0130.874
    Table 1. Data fusion and feature level fusion GANs models without normalized treatment
    特征提
    取方法
    归一化方法校正集检验集
    R2RMSECV偏差R2RMSECV偏差
    BiPLSSNV0.989
    0.989
    1.477
    1.467
    0.014
    0.055
    0.973
    0.973
    1.811
    2.012
    0.156
    0.259
    Min-Max-Nor0.993
    0.994
    1.136
    1.069
    0.023
    0.031
    0.982
    0.985
    1.655
    1.814
    0.211
    0.185
    VN0.9871.6530.0630.9352.9280.851
    Table 2. Statistics of the result of GAN prediction model using different normalization methods
    最优模型分类校正集验证集加标回收率
    R2RMSECV偏差R2RMSEP偏差
    UV全波段0.944 12.8170.1230.894 13.7490.80190.2~121.6
    NIR全波段0.904 63.5220.1470.877 44.1100.92582.1~130.4
    UV+NIR(LLDF)融合0.994 71.858-0.0350.945 72.974-0.12893.2~110.4
    UV+NIR(MLDF)融合0.997 70.9760.0160.994 71.3250.05998.4~103.1
    Table 3. Evaluation parameters of different quantitative prediction models
    Ying CHEN, Yang-mei XU, Yuan-jian DI, Xing-ning CUI, Jie ZHANG, Xin-de ZHOU, Chun-yan XIAO, Shao-hua LI. COD Concentration Prediction Model Based on Multi-Spectral Data Fusion and GANs Algorithm[J]. Spectroscopy and Spectral Analysis, 2021, 41(1): 188
    Download Citation