• Laser & Optoelectronics Progress
  • Vol. 57, Issue 7, 073002 (2020)
Jun Hu, Yande Liu*, Xudong Sun, Bin Li, Jia Xu, and Aiguo Ouyang
Author Affiliations
  • School of Mechatronics Engineering, East China Jiaotong University, Nanchang, Jiangxi 330013, China
  • show less
    DOI: 10.3788/LOP57.073002 Cite this Article Set citation alerts
    Jun Hu, Yande Liu, Xudong Sun, Bin Li, Jia Xu, Aiguo Ouyang. Quantitative Determination of Benzoic Acid in Flour Based on Terahertz Time-Domain Spectroscopy and BPNN Model[J]. Laser & Optoelectronics Progress, 2020, 57(7): 073002 Copy Citation Text show less
    Topology structure of BP neural network
    Fig. 1. Topology structure of BP neural network
    Terahertz absorption spectra of benzoic acid with different mass fraction in flour
    Fig. 2. Terahertz absorption spectra of benzoic acid with different mass fraction in flour
    Root-mean-square error of benzoic acid-doped flour changed with principal component factor
    Fig. 3. Root-mean-square error of benzoic acid-doped flour changed with principal component factor
    Regression coefficient of terahertz absorption spectrum of benzoic acid-doped flour in PLS model
    Fig. 4. Regression coefficient of terahertz absorption spectrum of benzoic acid-doped flour in PLS model
    Optimized results of BPNN for number of input vectors and hidden layers
    Fig. 5. Optimized results of BPNN for number of input vectors and hidden layers
    Relationship between predicted values and actual values of mass fraction in different models
    Fig. 6. Relationship between predicted values and actual values of mass fraction in different models
    Data setNumber ofsamplesMass fraction /%
    MinimumMaximumMeanStandard deviation
    Total1760.047919.9999.3326.243
    Calibration1320.047919.9999.3326.297
    Prediction440.047919.9999.3326.249
    Table 1. Mass fraction of benzoic acid in calibration and prediction sets
    Pretreatment methodPCrCRMSECrpRMSEP
    Original40.97220.01430.97470.0156
    Savitzky-Golay40.97190.01420.97460.0155
    MSC30.97580.01330.97550.0136
    Baseline correction50.97990.01240.96970.0143
    Normalization40.98060.01200.97900.0128
    Table 2. Results of terahertz absorption coefficient in PLS model by correction processing for flour doped with benzoic acid
    Type ofkernel functionEquationParameterNumber ofinput variablesrpRMSEPt /s
    LinK(xi,xj)=xixjγ= 12.6702640.9830.0126.162
    RBFK(x,xi)=exp(-x,xi2/2σ2)γ=16690σ2=229.4182640.9870.0114.368
    LinK(xi,xj)=xixjγ= 7.2903100.98020.01265.7252
    RBFK(x,xi)=exp(-x,xi2/2σ2)γ=219380σ2=3276.1100.98200.01284.8204
    Table 3. Results of terahertz spectral absorption coefficient of benzoic acid-doped flour samples in LS-SVM model
    Jun Hu, Yande Liu, Xudong Sun, Bin Li, Jia Xu, Aiguo Ouyang. Quantitative Determination of Benzoic Acid in Flour Based on Terahertz Time-Domain Spectroscopy and BPNN Model[J]. Laser & Optoelectronics Progress, 2020, 57(7): 073002
    Download Citation