The Research on Optimization of Near Infrared Traceability Model of Lycium Barbarum L.
|Course||Of Food Science|
|Keywords||Near-infrared fingerprints Lycium barbarum L. traceability quantify optimization|
As an authentic food, the quality of Lycium barbarum i.is closely related to the source of origin. The adulterated and counterfeited phenomenon of Lycium barbarum L. has become increasingly serious on the market at present. This behavior affected the geoherbalism of Lycium barbarum L. directly. Therefore, in order to better distinguish and identify Lycium barbarum L.in different origin, near infrared technology was used combined with chemical analysis method in this paper. The research object of this paper is the dried fruit of Lycium barbarum L.in in Ningxia Yinchuan、Ningxia Shi Zuishan、Ningxia Guyuan、Ningxia Zhongning、Inner Mongolia Linhe、Gansu Baiyin、Qinghai Xining、 Xinjiang Jinghe, sum to eight areas. Firstly, different chemical values of components of Lycium barbarum L. in different regions were collected and analysised. Then the origin traceability model and quantitative model of Lycium barbarum L. were established. The main results of the research as follows:(1) Collection and analysis of different chemical values of components of Lycium barbarum L. in different regions.The result showed that, the moisture content of-different Lycium barbarum L-area is in a same range through the drying processing, this can made the error of moisture on determination of various kinds of chemical composition of Lycium barbarum L.keep the same.The distribution frequency trends of content of total acid, total sugar, fat, protein, polysaccharide, flavonoids and betaine are obey the normal distribution, these content all reached as calibrating data requirements.(2) The optimization of Lycium barbarum L. traceability model of different origin with near infrared fingerprint spectrum. This part research was told that a traceability model of Lycium barbarum L. by using the near infrared spectra combined with cluster class independent soft model method (SIMICA) was established, and the qualities of traceability models which were established by the different spectral preprocessing methods were compared. The result showed that, in the950-1650nm wavelength range, after the original spectrum was treated by second order derivative, five smooth and SNV processing, it was seen the obvious characteristic absorption peak when its wavelength was in1135nm,1175nm,1235nm,1335nm,1395nm,1415nm,1535nm place;origin models included Xinjiang Jinghe, Ningxia Zhongning, Gansu Baiyin, Qinghai Xining, Ningxia Yinchuan, Ningxia Shi zuishan, Ningxia Guyuan and Inner Mongolia Linhe, when principal component number of these eight origin models is3, the traceability models of Lycium barbarum L.established by the SIMICA method were best; Under the10%significant level, the recognition rate of the unknown sample of eight origin was95%,85%,95%,95%,80%,80%,95%and95%; the misjudgement rate was2.86%,14.28%,2.86%,0%,5.72%,17.13%,0%and2.86%.(3) The optimization of Lycium barbarum L. chemical quantitative model of different origin with near infrared fingerprint spectrum. It was determined the content of medlar total acid, total sugar, fat, protein, polysaccharide, flavonoids, betaine of the eight different origin through chemical methods, and principal component regression (PCR) and partial least squares (PLS) model were established between near infrared spectral. information by using different methods and various chemical composition of Lycium barbarum L.The result showed that, these quantitative PCR models of chemical compositions in Lycium barbarum L. under the treatments of second derivative and five smoothing and SNV processing are the best. In these models of different composition, the calibrating correlation coefficient Re of total acid, total sugar, fat, protein, polysaccharide, flavonoids and betaine of Lycium barbarum L. were0.8333,0.8208,0.8228,0.8307,0.8311,0.8481,08491, cross-validation correlation coefficient Rv were0.8315,0.7906,0.8216,0.8264,0.8406,0.8377,0.8394.The second is the PCR model by using first derivative and five smoothing and SNV processing, the worst is the PCR model by using first derivative and five smoothing processing. The relevance and accuracy of PLS model of Lycium barbarum L. through different methods was obviously higher than relevance and accuracy of PCR model through different methods. These quantitative PLS models of chemical composition in Lycium barbarum L under the treatments of first derivative and five smoothing and SNV processing are the best. In these models of different composition, the number of principal components of total acid, total sugar, fat, protein, polysaccharide, flavonoids and betaine of Lycium barbarum L. were10,8,10,7,9,8,13; slope value of calibration curve were0.9031,0.8993,0.8983,0.8906,0.9053,09021,0.9036; the correlation coefficients Rc of calibration sets were0.9084,0.9012,0.9083,0.9166,0.9253,0.9101,0.9036; Slope value of cross validation curve were0.8945,0.8922,0.8934,0.8890,0.9029,0.8981,0.8998; the Cross-validation correlation coefficient Rv were0.9047,0.8917,0.8964,0.9093,0.9042,0.9058,0.8971.