| In recent years,China's food safety governance framework has become increasingly perfect,but food authenticity is still a concern,food authenticity traceability technology is in urgent need of development,technical problems need to be breakthrough,at present,food authenticity traceability technology and methods have not formed standards.Improving the origin traceability technology of near-infrared spectroscopy and the identification rate of mineral element origin traceability technology is the focus of research at home and abroad,especially for small-scale source samples.This experiment took 120 mung beans from Baicheng City in Jilin Province,Sishui County in Shandong Province,Tailai County in Heilongjiang Province and Dumeng County in Heilongjiang Province in 2018 and 2019 as the research objects.Fourier diffuse reflectance near-infrared spectrometer was used to scan the samples,combined with principal component analysis and partial least square discriminant analysis(PLS-DA)to screen out the near-infrared spectroscopy traceable fingerprint information related to the origin,and perform near-infrared analysis on mung beans from different origins.Research on the identification of the origin traceability of the spectrum.In order to screen the characteristic factors of the origin of mineral elements related to the origin,the inductively coupled plasma mass spectrometer(ICP-MS)was used to determine the content of 52 mineral elements in the sample.Through one-way ANOVA,the difference between the year,variety and origin The interaction analysis screens out the characteristic mineral elements directly related to the place of production,combines principal component analysis(PCA)and discriminate analysis(Linear Discriminate Analysis,LDA)to verify the traceability of the selected characteristic mineral elements,and explore the samples The influence of the origin distance and the number of samples on the origin traceability results.In order to further improve the accuracy of origin traceability,the extracted near-infrared spectrum fingerprints and mineral element characteristic factors were data fused,and a partial least squares discriminant analysis model was established to discriminate the origin of mung bean.The results of the study are as follows:(1)In terms of fingerprint extraction and identification of the origin traceability of mung bean near infrared spectroscopy:By scanning the near-infrared spectra of mung bean from 4 producing areas,using principal component analysis to extract characteristic wavelengths,the cumulative contribution rate of the first three principal components reached 92.17%.The fingerprint information of the near-infrared spectra contained in the three principal components was screened out.In the quantitative analysis,by comparing the accuracy of the model established by 5 pretreatment methods,it is concluded that the partial least squares discriminant analysis model established by the vector normalization+multivariate scattering correction(MSC)spectral pretreatment method is the most stable.The overall discrimination rate of the four origins is 90%,and the correct discrimination rates of the four origins are 100%in Baicheng,80%in Dumeng,80%in Tailai,and 100%in Surabaya.It shows that the near-infrared spectroscopy traceability technology is feasible for mung bean origin traceability,but the accuracy of mung bean discrimination in the small distance range of Du Meng and Tailai needs to be further improved.(2)In the research on traceability fingerprint screening and identification of the origin of mung bean mineral elements:Through the analysis of the single-factor variance analysis of the mineral element content of mung bean in the 4 producing areas,the interaction between the year,variety and producing area,and the analysis of the variance contribution rate of each factor,the elements P,K,Mg,As,Na,B,Ba in mung bean are obtained.The content of Mn,Co,Se,Cs,Pb,Ca,Sr,Cu,Zn,Sn,Lu,Cr,Pt,Pd is closely related to the place of production,and the content of Al,Ni,Cd is closely related to the year;Fe,Rb The content of,Mo,Ag,and La is closely related to the variety.Principal component analysis and Fisher discriminant analysis were used to screen the 21 characteristic mineral elements directly related to the production area.The overall discrimination rate of the 4 production areas was 96.7%,the cross-validation discrimination rate was 93.3%,and the original correct discrimination rate of the 4 production areas.And cross-validation discrimination rates are 100%and 96.7%in Baicheng,90%and 83.3%in Dumont,96.7%and 93.3%in Tailai,and 100%and 100%in Surabaya.It shows that the selected mineral element producing area characteristic factors can distinguish the mung beans from 4 producing areas,but the discrimination rate for small-scale traceability samples needs to be improved.Therefore,the above research shows that the traceability range of the sample is an important factor affecting the traceability results.By studying the influence of the distance of the sample origin and the number of samples on the traceability result,it is concluded that the greater the distance of the origin of the sample,the higher the discrimination accuracy,and the larger the number of samples,the higher the discrimination accuracy.(3)In terms of the fusion of characteristic near-infrared spectroscopy and mineral element information,the traceability of mung beans from different origins:Combine the extracted 3 near-infrared spectroscopy traceability characteristic fingerprints and the selected 21 mineral element characteristic factors directly related to the place of origin.Combine 120 samples from 4 places of origin to obtain a 24×120 data matrix,which is analyzed by partial least squares discriminant analysis,The overall discrimination rate of the 4 mung bean producing areas reached 100%.Compared with the single technology origin traceability model,the fusion model has better applicability and accuracy for discrimination.Effectively improve the effect of identifying the origin of small-scale samples. |
PDF Full Text Request