| The papers studied the rapid detection of water, fat, protein and total sugar content inGuoKwei being based on fourier transform near infrared spectroscopy, and selected the bestmodeling parameters by comparing different modeling parameters on the effect of the model.The specific research results were as follows.National standard methods as the main reference methods, the moisture, fat, protein andtotal sugar content range were13.97%～34.58%,0.18%～16.77%,7.8%～11.61%and1.09%～4.59%,respectively.The average value were23.35%,7.13%,9.47%and2.71%,respectively. The various components of the sample range covers essentially the content thatmay appear in GuoKwei. The samples have a certain representation.The near infrared spectrum analysis model of moisture, fat, protein and total sugar wereconstructed. The original spectra of GuoKwei were obtained, partial least squares is used. TheR2value of the near infrared spectrum analysis model of moisture, fat, protein and total sugarwere98.07%,96.53%,95.49%and73.12%, and its RMSECV were0.527%,0.786%,0.215%and0.452%. The R2value of validation set were96.11%,97.57%,96.20%and74.69%, andRMSEP were0.692%、0.653%、0.202%and0.448%,respectivly. Internal and externalcross-certification test have shown that calibration models have good prediction performanceexcept the model of total sugar and they could meet the accuracy needs of prediction ofmoisture, fat and protein content of GuoKwei. |
PDF Full Text Request