Chemical Quality And Fingerprints Of Yunan Flue-Cured Tobacco And Their Application In Quality Classification

Improving the quality stability of blend threshing plays a key role in building and developing of tobacco brand. In order to meet the different demand of leaf for Tobacco Company, one more scientifical and rational method to evaluate and classify tobacco quality should be chosen. This research analyzed the chemical composition of tobacco from 12 tobacco-producing areas in Yunan during 2009 to 2010, established the fingerprint and databas, and developed an appropriate method to classify and evaluate tobacco quality. The main research contents included:(1) The variance analysis of conventional chemical compositions of tobacco among different producing areas in Yunan. (2) The variance analysis of conventional chemical compositions among process re-grading. (3) Determination of the optimal extractant for establishing the tobacco fingerprint and developing the standardization procurement procedure. (4) Correlation analysis between the fingerprint and conventional chemical compositions. (5) Applying for the flue-cured tobacco quality classification in Yunan by using the fingerprint. (6)The establishment and application of tobacco style type by using the fingerprint. The main results are as follows:1. The fluctuation of the chemical composition of tobacco leaves was stable from different producing areas in Yunan, it provided the objective condition for the quality classification and the blend threshing. The chemical composition ranges about pH, nicotine, total water soluble sugar, reduced sugar,non-reduced sugar, the ratio of sugar and nicotine of grade C3F tobacco leaves from the producing area in yunan in 2009 were: 5.32～5.75,1.27%～3.92%,20.86%～36.15%,20.18%～32.03%,0.63%～11.69%,5.95～28.52; Grade B2F tobacco leaves in 2009:5.19～5.53.2.50%～4.00%,17.53%～33.64%,13.55%～27.65%,1.13%～9.24%,4.14～13.01; Grade X2F tobacco leaves in 2009:5.38～5.71,1.02%～2.07%,22.10%～37.48%,19.31%～29.87%,1.70%～10.06%,13.41～36.92. Grade C3F tobacco leaves in 2010:5.25～5.67,1.11%～3.84%,20.20%～37.94%,13.72%～31.81%,1.85%～10.42%,5.55～34.05; Grade B2F tobacco leaves in 2010:5.17～5.50,1.84%～4.86%,13.76%～34.32%,12.50%～27.90%,0.8%～10.76%,2.83～18.62; Grade X2F tobacco leaves in 2010:5.37～5.65,1.16%～2.88%,17.81%～45.29%,15.22%～31.60%,0.92%～13.96%,6.54～30.72.2. A HPLC method is used to establish flue-cured tobacco fingerprint. Specially including:(1) Developed the best extraction method, (2) Developed the chromatographic condition of separation of the flue-cured tobacco, the RSD of accuracy, repeatability and stability of the total peaks area is <6%, The similarity values of 7 batches of samples is more than 0.98.3. The time-sharing database of fingerprint was innovative established for showing the characteristics of different tobacco leaves fingerprint. In this way, it can effective sort the amounts of data and reduce workload and error.4. The principal component analysis showed, eleven fingerprint time sharing peaks were extracted as three main principal components, which the cumulative contribution reached to 91.5%. The correlation analysis was established between the nicotine, total water soluble sugar, the ratio of sugar and nicotine and three main principal components. The nicotine content had a significantly positive correlation with the second and third principal component, the total water soluble sugar had a significantly negative correlation with the first and second principal component; the ratio of sugar and nicotine had a significantly negative correlation with the first and second principal component, and a significantly positive correlation with the third principal component.5. The multiple regression models were established between the values of the nicotine, total water soluble sugar and the ratio of sugar and nicotine with the fingerprint time sharing peaks(1)the optimal nicotine regression model:Y=0.00051615X2-0.44122(X2:the areas of peaks in 3-6min); F=3227.64, Pro.=0.01.(2)the optimal total water soluble sugar regression model:Y=-0.0536X1-0.0013X2-0.03472X8+0.02099X11+47.01802 (X1:the areas of peaks in 0-3min,X2:the areas of peaks in 3-6min,X8:the areas of peaks in 48-51min,X11:the areas of peaks in 57-60min); F=14.21, Pro.=0.01.(3)the optimal the ratio of sugar and nicotine regression model:Y=-0.00328X2+0.0059X4-0.01253X7+31.36150(X2:the areas of peaks in 3-6min, X4: the areas of peaks in 21-24min, X7:the areas of peaks in 42-45min); F=81.69, Pro.=0.01.6. The cluster analysis was applied for the tobacco quality classification from different producing areas in yunan by the fingerprint, the result showed,In 2009,the C3F level samples were grouped into four types,the B2F level samples were grouped into three types,the X2F level samples were grouped into two types.In 2010, the C3F level samples were grouped into three types,the B2F level samples were grouped into five types,the X2F level samples were grouped into three types.7. The regression model could accurately predict the chemical composition of tobacco by the fingerprint peaks. The result showed, the CV values of nicotine content,total water soluble sugar and the ratio of sugar and nicotine prediction were under 0.05,0.13,0.25. The neural network model can predict the type style of tobacco leaf quality and evaluated on the leaf.8. The neural network prediction model was established for the quality and classification of tobacco leaves based on fingerprint database. The result showed, the model could accurately predict the type style of tobacco leaf and the evaluation of tobacco quality.