Standard Free Transfer of Near Infrared Spectral Model by Two-step Wavelength Screening Method Based on MC£­UVE£­VIP Algorithm

KeyWord:two-step wavelength screening  Monte Carlo-uninformative variable elimination£¨MC£­UVE£©  variable importance in projection£¨VIP£©  model transfer without standard samples  prediction of corn protein

Abstract:

      Previous studies have shown that selecting suitable wavelengths to establish a near-infrared£¨NIR£© spectral model can realize the transfer of calibration model without standard samples£®Using the corn spectra of three near-infrared instruments and the protein in corn as the case study£¬a two-step wavelength selection method of MC£­UVE£­VIP is proposed in this paper by combining Monte Carlo-uninformative variable elimination£¨MC£­UVE£© method with variable importance in projection£¨VIP£© method£®The method selects important and informative wavelength variables to build partial least square£¨PLS£© calibration model for protein in corn£®Firstly£¬the method uses MC£­UVE to screen the set of UUVE consisting of informative wavelengths£¬whose stability parameter for screening wavelengths are greater than a threshold£¨Mthreshold£©£®Then£¬the VIP algorithm is used to screen the important and informative wavelength set of UUVE-VIP£¬whose VIP parameter is greater than the average value of VIP of all wavelengths in the UUVE£®A PLS near-infrared spectral model for predicting protein in corn is established based on UUVE-VIP£®The number of latent variables of the PLS model is determined according to the principle that accumulative contribution rate of the first latent variables should be higher or equal to 99.9%£®The model is of few variables£¬robust£¬good at interpretability£¬and fast at calculation£®The mean absolute relative error£¨MARE£© of predicting protein in the samples tested on the secondaries is 1.64% and 1.88%£¬respectively£®While the MARE of the secondary samples predicted by MC£­UVE and VIP models are 5.19%£­5.40% and 6.23%£­7.16%£¬respectively£®Therefore£¬the near-infrared spectral model of corn protein based on the MC£­UVE£­VIP two-step wavelength screening method can be directly transferred to the secondary instruments without correcting model or secondary samples spectra£®The transfer performance of the model is better than that of the model based on single MC£­UVE or single VIP wavelength selection algorithm£¬and is better than full wavelength model£®The adoptability of the two-step wavelength selection method of MC-UVE£­VIP for the other data set needs to be inspected further.