Near Infrared Instrumentation And Researches On Foodstuff Component Calibration Technology

Near-Infrared Spectra analysis techniques based on the relativity between sample's compositions and Near-Infrared spectra is used for qualitative and quantitative measurement for sample's compositions. It includes near infrared instrumentation, spectra data processing software and calibration sample. It is a green analysis technique for its very rapid, non-invasive and no chemical reagent characteristic.In this paper, fundamentals of the interaction of radiation with matter, basic structure of common near infrared instrumentation and near infrared calibration was introduced firstly. The basic structure of near infrared reflective instrumentation for corn quality analyzing was introduced mainly. The instrumentation is imbursed and manufactured with outlay of Nationality the Tenth Five Year Science and Technology Plan of Tackle Key Problem. We use the instrumentation to analyze moisture of corn sample and protein of flour sample. Good calibration model was got. Secondly, the relation between wavelength number and the calibration of flour sample and corn sample was analyzed, and the experiment shows that adding calibration wavelength number will cause over combine when the number cover the spectra information of sample composition. So, when the wavelength number of flour sample exceeds 2 and the wavelength number of corn sample exceeds 3, their actual validation ability reduces. Thirdly, in order to fine the reference wavelength of better calibration model, use moisture characteristic wavelength and other wavelength to build the calibration model respectively, and validate it with cross-validation. The result shows that the reference wavelengths of corn powder and flour are different. Because different kinds of sample have different chemical structure, and some chemical bonds will influence the absorption of the moisture. Analyzing flour protein quality, the calibration model is the best when the wavelength number receives 7. In order to decrease regression wavelength number and regression coefficient value, a subtraction operation methodof originality spectrum was introduced, and the regression model built in this method is better than others, and RMSECV reduce. And lastly, discussing the effect of multiplicative scatter correction (MSC) on the one wavelength calibration model, and comparing the calibration model of one character wavelength and that after MSC, the result shows that RMSECV of moisture character wavelength calibration model after MSC reduces observably, but the RMSECV of protein calibration model can not reduce evidently. Because the RMSECV reduce, it can make out that MSC can remove the effect of disturbing factor.