| Cereals are the main body of traditional diet in most countries in the world.Due to the similarity of grains,more expensive grains may be mixed with cheaper grains.In certain cases,multiple cereal flours may be mixed for applications.Therefore,for safety,economic,regulatory,functional or nutritional reasons,it is necessary to determine the composition of different cereal flour mixtures,as well as the adulteration of cereal flours.Front-face synchronous fluorescence spectroscopy（FFSFS）adopts the mode of synchronous scanning excitation and emission spectroscopy,which greatly simplifies the spectrum.It directly measuring the fluorescence excitation and emission on the sample surface,avoiding the inner filtering.Thus,it reduces spectral overlap and scattered light interference.This method is simple,sensitive,fast,and non-destructive,and is increasingly used in the quantification,adulteration detection and type identification of various foods.At present,the fluorescence characteristics of many grain flours are well documented.However,all these documents focus on the qualitative classification or analysis of nutrients based on plant sources.In contrast,the accurate quantitative detection and composition analysis of mixed flour by fluorescence spectroscopy remains a difficult challenge.In this dissertation,several grain flour is selected as the research object.Their front-face fluorescence spectra were studied and analyzed,and then combined with chemometric methods to analyze the adulteration of maize flour or soybean flour in quinoa flour.Fast,nondestructive determination and composition analysis and quantitative identification of maize flour,sorghum flour,barley flour,wheat flour and other grain mixtures was also achieved.This dissertation is mainly divided into the following two aspects:（1）FFSFS was used for qualitative discrimination and quantitative analysis of quinoa flour adulteration with maize flour or soybean flour.This experiment is mainly divided into two cases:（i）quinoa flour was adulterated with maize flour or soybean flour individually;and（ii）maize flour and soybean flour（6-50 wt %）were simultaneously adulterated in quinoa flour.For single adulteration of maize flour or soybean flour,principal component analysis（PCA）and linear discriminant analysis（LDA）were used to make preliminary qualitative discrimination,and most of the unadulterated quinoa flour was separated from the adulterated samples.In addition,the two adulterated classes show clear separation with relatively tight“in-group” cluster.Then the prediction models were constructed based on the combination of unfolded total synchronous fluorescence spectra and partial least square regression（PLSR）,and were validated by five-fold cross-validation and external validation.The coefficient of determination（R2）were in the range of 0.939-0.967,and the highest root mean square error（RMSE）were 3.3%.For dual adulterants,a general PLS2 model for the simultaneous determination of maize and soybean flours was developed,which produced suitable results,with the determination coefficient of prediction（R2p）> 0.9,root mean square error of prediction（RMSEP）< 5% and residual predictive deviation（RPD）> 3.The limits of detection（LODs）were 9.0% and 6.9% for maize flour and soybean flour,respectively.Although the PLS2 results were not as satisfactory as those of PLS1,it was acceptable for reliable routine analysis.Although the results of PLS2 were not as satisfactory as PLS1,they were acceptable for reliable routine analysis.Furthermore,most spike recoveries were in the range of 80%–120%.（2）Composition analysis of various cereal flour blends by FFSFS.Maize and sorghum as well as wheat and barley are closely related crops of the same family.They share many similarities and can be mixed in certain applications.However,in some cases,one grain may be preferable to another for various considerations.Hence the determination of the composition of these binary blends may be necessary for economic,regulatory or nutritional reasons.Furthermore,quaternary blends composed of maize,sorghum,wheat and barley flours were tested to assess the capacity of the strategy.Prediction models were constructed by PLSR for both the two cases,respectively.For the binary blends of maize and sorghum or wheat and barley flours,the models produced suitable results.The R2 values of calibration and five-fold cross-validation were all larger than 0.96.For quaternary blends of the four flours,the PLSR model simultaneously predicted the contents of four flours with acceptable results,with the R2 values in the range of 0.896–0.963,RMSEP < 8.5% and RPD > 3.Additionally,the spike recoveries of most binary blends ranged in 90-110%.Although the spike recoveries of quaternary blends were not as good as that of binary blends,most of the them belonged to the range of 80%–120%.Finally,both the intra-day repeatability and interday reproducibility were evaluated by coefficients of variation（CVs）and were all less than 30%. |
