Based On The Vector Subspace Angle Criterion Analysis Food And Drug Additives

The trend of modern analytical chemistry is the fast, direct, and quantitative analysisfor the complex and multi component chemical system. That "Mathematical separation"partially or completely replace the "physical chemical separation" is the Characteristics ofChemometrics. Vector-subspace angle is a subspace judgment. Base on it, the quantitativemethod of one-dimensional complex spectral is proposed. Robust criterion is theneccesary condition of robust chemometrics method. The judgment based oncorresponding subspace of system was robuster than the judgment based on characteristicsof samples matrix. The instruction of multi-wavelength spectral libraries of samplebackground is a key of the Vector-subspace angle Criterion. For different chemicalsystems, at first, through Angle-subspace algorithm combining with background database,establish a accurate rapid quantitative analysis method of the ingredient of mixed system,of which results were accurate compared with HPLC. Furthermore, the multi-wavelengthspectral libraries of sample background which the determinated compounds were isolatedfrom sample by HPLC-spectrometer were constructed and it needn't repeated collectiononce it had been constructed. So that the quantitative analysis of one-dimensional spectrabasis on the Vector-subspace angle criterion can be come true, which it needn't correctiondepended on other instruments and it can simplify the quantitative analysis of enormoushigh-order signals generated by the modern analytical instruments. Studies focused on themethodologies and applications of the Vector-subspace angle criterion.(1) In Chapter3, the analysis of sodium benzoate in orange juice was determined bymixing background components in simulation cases, which was basis on vector-subspaceangle criterion. In Chapter4, the analysis of benzoic and salicylic in compound benzoicacid tincture was also verified. For the simple and difinite samples, the multi-wavelengthspectral libraries of sample background can be collected on spectrometer by mixingbackground components. And multi-wavelength spectral libraries of the samples andstandard components with least-square regression were constructed to quantitativeanalysis basis on vertor-subspace angle criterion. The results showed that: comparing theresult of Vector-space angle criterion with the result of HPLC, the relative error of sodiumbenzoate in orange juice was less than5.0%and the recories were99.20%-103.73%. Therelative error of benzoic and salicylic in compound benzoic acid tincture were respectivelyless than3.75%and4.16%, and the recoveries of benzoic and salicylic were respectively 98.33%-102.00%and98.67%-104.00%.(2) In Chapter5, the multi-wavelength spectral libraries of sauces background werecollecte on spectrometer through celectting sauces in which there were no sorbic acid(potassium) by HPLC. And the multi-wavelength spectral libraries of the sauces andsorbic acid (potassium) standard with least-square regression were constructed toquantitative analysis of sorbic acid (potassium) in sauces basised on vertor-subspace anglecriterion. The results showed that: comparing the extraction results of ether and dilution,the relative error of sorbic acid (potassium) in sauces was less than4.16%and the recorieswere98.00%-102.00%by from ether. The relative error of sorbic acid (potassium) insauces was less than4.5%and the recories were95.44%-102.64%in the sauces dilution.So it was proved that the quantitative analysis in multi components based on the vector-subspace angle criterion can be determinted derectly and quickly with no complexpretreatment.(3) In Chapter6, the quantitative analysis of tartrazine in functional drink of wasestabilished with Spectra-chromatographic based on the vector-subspace angle criterion.Owing to the vintamines, minerals, pigment in drink, the drink background are toocomplex to mix them completely. Then, the multi-wavelength spectral libraries offunctional drink background which the tartrazine was isolated from functional drink bySpectra-chromatographic were available. Because the high-order matrix of backgroundspectral libraries were too enormous to computer, so that the dimension reduction can berealize by principal component analysis and singular value decomposition with the Matlab.The one-dimensional spectra of tartrazine standard and functional drink and the spectrallibraries of functional drink were computed to determinate the tartrazine in functionaldrink by the algorithm of vector-subspace angle. Comparing the result of Vector-spaceangle criterion with the result of HPLC, the relative error of tartrazine was less than5.00%and the recories were96.80%-99.00%.