Acta Optica Sinica, Volume. 41, Issue 22, 2230001(2021)

Multi-scene Optimization Method of Water COD Measurement Based on Direct Spectroscopy

Enhua Liu, Decao Wu*, Jie Wang, Wanyao Tan, Junhao Pu, Binbin Luo**, Bin Tang, Zourong Long, and Mingfu Zhao
Author Affiliations
  • Chongqing Key Laboratory of Optical Fiber Sensor and Photoelectric Detection, Chongqing University of Technology, Chongqing 400054, China
  • show less
    Figures & Tables(10)
    Turbidity direct fitting correction results based on modified power function. (a) Fitting results of the particles’ simulation spectra by modified power function with particle size of 0.1-100 μm; (b) turbidity direct fitting correction results of diatomite sample spectra
    Spectra of samples with different COD concentrations in two scenes. (a) Original spectra;(b) absorbance normalized spectra
    Flow chart of multi-scene optimized COD measurement method
    Schematic of water sample spectrum measurement system
    Scene matching results of the sample spectra in the test set. (a) Scene matching rate of the sample spectra;(b) Cmax of the sample spectra
    Scenes matching results of test set based on three typical classification methods. (a) SVM; (b) BP-ANN; (c) K-NN
    Influence of three turbidity characteristic fitting functions on SLCSM-PLS COD calculation
    • Table 1. Accuracy statistics of four scene-matching methods

      View table

      Table 1. Accuracy statistics of four scene-matching methods

      MethodAccuracy /%
      BP-ANN55.56
      K-NN62.22
      SLCSM100.00
      SVM91.11
    • Table 2. Experimental results of multi-scene COD calculation

      View table

      Table 2. Experimental results of multi-scene COD calculation

      Scene labelActual value /(mg·L-1)Holistic PLS modelingSVM-PLS modelingSLCSM-PLS modeling
      Calculation value /(mg·L-1)Error rate /%Calculation value /(mg·L-1)Error rate /%Calculation value /(mg·L-1)Error rate /%
      180.441175.931361.8779.081.6979.081.69
      2141.18130.467.59140.750.30140.750.30
      337.78545.591344.12264.43599.9236.593.15
      4160.881573.13877.83160.810.04160.810.04
      5109.93862.32684.43111.141.10111.141.10
      63.63220.675979.0640.721021.763.640.28
      720.01285.891329.4519.691.6019.691.60
      86.03156.082488.395.911.995.911.99
      96.1275.151127.9423.82289.226.021.63
      1010.31100.77877.4019.2486.6110.102.04
      Average execution time /s0.3630.5530.762
    • Table 3. Statistics of the influence of three turbidity characteristic fitting functions on the error rate of COD calculation in SLCSM-PLS method

      View table

      Table 3. Statistics of the influence of three turbidity characteristic fitting functions on the error rate of COD calculation in SLCSM-PLS method

      Scene labelError rate /%
      Power functionExponential functionQuadratic polynomial
      11.6911.4314.06
      20.3010.4940.57
      33.1513.2412.25
      40.048.459.61
      51.1054.7365.76
      60.282.932.09
      71.6010.279.53
      81.999.628.98
      91.632.342.29
      102.042.953.17
      Average error rate1.3612.6516.83
    Tools

    Get Citation

    Copy Citation Text

    Enhua Liu, Decao Wu, Jie Wang, Wanyao Tan, Junhao Pu, Binbin Luo, Bin Tang, Zourong Long, Mingfu Zhao. Multi-scene Optimization Method of Water COD Measurement Based on Direct Spectroscopy[J]. Acta Optica Sinica, 2021, 41(22): 2230001

    Download Citation

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

    Category: Spectroscopy

    Received: Apr. 19, 2021

    Accepted: Jun. 1, 2021

    Published Online: Oct. 29, 2021

    The Author Email: Wu Decao (wudecao@163.com), Luo Binbin (luobinbin@cqut.edu.cn)

    DOI:10.3788/AOS202141.2230001

    Topics