| Baijiu market is complex and huge.In order to maintain the good order of baijiu market,effective measures should be taken to strengthen the supervision and management of liquor market.At present,sensory evaluation is the gold standard for baijiu detection.However,training of baijiu tasters is difficult,and the testing process is easy to be affected by their own physical and psychological and cannot be fully data-based and standardized.Other precision large-scale instruments for qualitative and quantitative analysis for baijiu have some limitations in identification of baijiu flavor type and brand due to complex pretreatment,long detection time and complex operation.According to combination of sensor array sensing technology and nano materials,we successfully identification 16 kinds of baijiu by constructing 5 targeted nano optical sensor arrays.This is the first time that the sensor array has been designed for baijiu identification via principal components analysis.RBFNN is widely used in pattern recognition,automatic control,fault diagnosis and other fields because of its strong nonlinear fitting ability.In order to improve the shortcomings of a single component sensitive sensor array in blind sample test,identification with other 5 distilled baijiu,baijiu with non-baijiu determination,we integrated 5 sensitive sensor arrays,and used RBFNN and LDA to deeply mine the data.Based on RBFNN,which has high accuracy in blind sample testing,a baijiu Test App was developed for accurate identification of baijiu.This study systematically demonstrated the possibility of nanomaterials-based sensor array in the detection of single substances,the identification of distilled spirits and the determination of baijiu with non-baijiu samples,which laid the foundation for the practical application of sensor array detection technology in the field of baijiu detection.The specific conclusions are as following:（1）One organic acid sensitive QDs-based fluorescence sensor array was successfully constructed for qualitative and quantitative detection of 8 kinds of organic acids and the identification of 16 kinds of baijius.Fluorescence spectra and TEM were used to characterize QDs and explain the reaction mechanism.Optimal fluorescence emission peaks before and after reaction were normalized for subsequent pattern recognition and fingerprint map establishment.In the identification and detection of organic acids,provided sensor array can correctly identify 8 kinds of organic acids with certain quantitative detection ability and good selectivity.On this basis,by using of fingerprint map and pattern recognition analysis,provided sensor array has been successfully applied to identify 16 kinds of baijius.Good universality and practical application potential could be proved by the high LOO accuracy of subsequent blind sample test（87.5%）.The multi-dimensional regression analysis of sensor array response Euclidean distance with total acid content（GB/T 10345-2007）of 8 baijius showed a significant correlation and good fitting results of these 2 factors（P<0.01,R2≥0.778）.Baijiu,brandy,rum,whisky,vodka and gin are referred to as the world’s 6 major distilled spirits.It is of great significance to identify baijiu from other distilled spirit.In the identification 16 baijius from other 5 distilled spirits,provided sensor array can identify 21 kinds of distilled spirits one by one,but cannot determine baijiu samples with non-baijiu samples.（2）One Ag NPRs-based colorimetric sensor array was successfully constructed for qualitative and quantitative detection of 8 thiol compounds and the identification of 16baijius.Principle is the case that positively charged metal ions(Cd2+,Fe2+,Fe3+,Cr3+)can spontaneously compete for the binding sites between Ag NPRs and thiol containing substances,reducing the anti-etching ability of Ag NPRs under acidic conditions.UV Vis absorption spectroscopy and TEM were utilized to investigate the morphology change and color change of Ag NPRs.ΔRGB value in red,green,blue channels were used for subsequent pattern recognition and fingerprint map establishment.In the identification and detection of sulfhydryl containing substances,provided sensor array can correctly identify 8 kinds of sulfhydryl containing substances with certain quantitative detection ability and good selectivity.The etching processes of Ag NPRs in baijiu were studied by TEM-EDS.On this basis,by using of fingerprint map and pattern recognition analysis,provided sensor array has been successfully applied to identify 16kinds of baijius.Good universality and practical application potential of this sensor array could be demonstrated by the high LOO accuracy of subsequent blind sample test（93.8%）.In the identification of 16 baijius and other 5 distilled spirits,the addition of distilled spirit samples led to the confusion of the model.This array could not identify21 kinds of distilled spirits nor could determine the baijiu samples with non-baijiu samples.（3）One Au NPs-based colorimetric array on was constructed by controlling the loading of Ag+on 2 sizes Au NPs.It was successfully applied in the qualitative and quantitative detection of 7 aldehydes and the identification of 16 kinds of baijius.Ag+attracted to Au NPs by positive and negative charges can be reduced by aldehyde group to form a core-shell structure of Ag@Au NPs,producing a dramatic color change.The relationship between Ag+reduction process with Au NPs charge and particle size were analyzed by UV Vis absorption spectrum,TEM and zeta potential.TheΔRGB values were used for subsequent pattern recognition and fingerprint map establishment.In the identification and detection of aldehydes,provided sensor array can correctly identify 7aldehydes with certain quantitative detection ability and good selectivity.On this basis,via using fingerprint map and pattern recognition,this array was successfully applied to identify 16 kinds of baijius.Good universality and practical application potential could be proved by the high LOO accuracy of subsequent blind sample test（90.6%）.The multi-dimensional regression analysis of sensor array response Euclidean distance with total aldehydes content（QB/T 1326.9-1991）of 8 baijius showed a significant correlation and good fitting results of these 2 factors（P≤0.01,R2≥0.803）.In addition,the situation that baijiu may inhibit sensor array response leading to low fitting degree was analyzed.In the identification test between 16 baijius and other 5 distilled spirits,this array can not only identify 21 kinds of distilled spirits,but also can determine baijiu samples with non-baijiu samples.（4）One Tb/Eu-MOF fluorescence sensor array was successfully constructed for qualitative and quantitative detection of 15 kinds of small molecular organic compounds and the identification and recognition of 16 kinds of baijius.Different guest molecules can produce different host guest effects with the Tb/Eu-MOF,change the energy conversion pathway and transition behavior between metal framework and organic ligand.Fluorescence spectra and TEM-EDS were used to characterize the synthesized Tb/Eu-MOF materials.6 fluorescence peaks of Tb/Eu-MOF 1 before and after the reaction were normalized for pattern recognition and fingerprint map construction.In the identification and detection of small molecular organic compounds,provided sensor array can correctly identify 15 kinds of small molecular organics with certain quantitative detection ability.On this basis,by means of fingerprint and pattern recognition analysis,this array has been successfully applied to identify 16 kinds of baijius.Good universality and practical application potential could be proved by the high LOO accuracy of subsequent blind sample test（96.9%）.In the identification test of16 baijius and other 5 distilled spirits,although this array can identify 21 distilled spirits,it cannot realize the determination of baijiu samples with non-baijiu samples.（5）One Au NRs-based colorimetric sensor array was successfully constructed for qualitative and quantitative detection of 20 kinds of reducing substances and the identification of 16 kinds of baijius via reducing substances can reduce the etching degree of oxidized ox-OPD on Au NRs.UV Vis absorption spectra and TEM-EDS were utilized to characterize the morphology of Au NRs before and after etching.TheΔRGB value of the reaction was used for subsequent pattern recognition and fingerprint map establishment.In the identification and detection of reducing substances,provided sensor array can correctly identify 20 kinds of small molecular organic compounds with certain quantitative detection ability.On this basis,by means of fingerprint map and pattern recognition analysis,provided sensor array has been successfully applied to identify 16 kinds of baijiu.Good universality and practical application potential could be proved by the high LOO accuracy of subsequent blind sample test（90.6%）.Potassium permanganate index（GB 11892-89）,an indicator for the detection of oxidizable substances in drinking water,was used to quantitatively detect the total content of reducing substances in baijiu.Multi-dimensional regression analysis of sensor array response Euclidean distance with potassium permanganate index of 8baijius showed a significant correlation and great fitting results of these 2 factors（P<0.01,R2≥0.849）.In the identification test of 16 baijius and other 5 distilled spirits,this array can successfully identify 21 distilled spirits,and can determine baijiu samples with non-baijiu samples.（6）Based on 5 targeted nanomaterials-based sensor arrays,a super large multi-dimensional integrated sensor array was constructed to improve the limitations of single targeted sensor array,such as the unsatisfactory blind samples test results,the identification errors of 21 kinds of distilled spirits,and the low determination accuracy of baijiu with non-baijiu.In this chapter,the LDA and newrb function of RBFNN were employed to evaluate the recognition performance of super large multi-dimensional synthetic sensor array for distilled spirits.In LDA results,the accuracy rate of LOO in blind sample test was 84.4%,and the accuracy rate of 21 distilled spirits identification and determining baijiu samples with non-baijiu samples were 100%.In RBFNN results,the accuracy rate of blind sample test was 93.8%,and the accuracy rate of 21 distilled spirits identification and determining baijiu samples with non-baijiu samples were100%.The above results show that the super large multi-dimensional comprehensive array can accurately identify 21 kinds of distilled spirits,and accurately determine baijiu samples with non-baijiu samples.The RBFNN newrb function,which performs well in blind sample test,was selected to construct Baijiu Test App for identifying of baijiu.In1120 times tests,the test accuracy of App was 96.875%,which proved that the app could accurately identify baijiu samples and owns good stability. |
