Research Of The Comprehensive Evaluation Methods On The Tobacco Varieties

The relationship between agronomic characters and economic characters and between chemical composition and sensory quality was analyzed by using the method of canonical correlation ,based on the test data of tobacco varieties in variety regional test in 2007 to primary construct index system of tobacco varieties in comprehensive evaluation. Meshing the advantage of optimization model and Radar chart, the comprehensive evaluation of varieties was formed by the research of weighting and evaluating. This method was validated by the test data of tobacco varieties in North China variety regional test in 2007 and 2008, and the results were in accordance with actual performance. The main results are as following:1. Agronomic characters of flue-cured tobacco had a significant correlation with the economic characters. The correlation between length-width ratio, length, width of middle leaves, leaf number and high and middle class leaf proportion, the average price, production value play a leading role in relationship of these variables(agronomic characters and economic characters). Economic characters can be predicted by length-width ratio, length of middle leaves. As constructing the index system, high and middle class leaf proportion, the average price, production value were erased appropriately, and choose yield per mu, high class leaf proportion.2. Chemical composition of flue-cured tobacco had a significant correlation with the sensory quality. The correlation between length-width Smoking aroma quality, irritancy and aftertaste and Total sugar, potassium and K /CL play a leading role in relationship of these variables(sensory quality and chemical composition). The sensory quality can be predicted by Total sugar, potassium, K /CL, sugar -nicotine ratio and reducing sugar. As constructing the index system,choose the chemical composition instead of the sensory quality ,because evaluating the sensory quality is hard.3. Indexes system of tobacco comprehensive evaluation of varieties are constructed primarily according to the results of canonical correlation and importance of indexes. Agronomic characters include stem length, leaf number and length of middle leaves. Economic characters include high class leaf proportion, the average price and yield per mu. Chemical composition include nicotine, potassium, Nitrogen--nicotine ratio, K /CL, sugar-Nitrogen ratio and Starch. The sensory quality includes smoking quantity, aroma quality and Concentration. Characters of disease resistance include TMV, CMV, PVY and the resistance of Brown spot, Leptosphaeria maculans, knot nematode and bacterial wilt. Choose the right disease indexes as evaluating the varieties.4. Make sure the weighting method. Construct two weighting methods according to the flexibility of optimization model and the character of tobacco. The first method is the Deviation maximization method, which reflects the difference influence of characters by the evaluating indexes. The second method is the one meshing the TOPSIS and the optimization model., which constructs the model aimed at testing varieties closing to the ideal variety. This method reflects the contributed proportion when the Euclidean distance bwteen testing variety and ideal variety is as small as possible.5. Construct the comprehensive evaluation of tobacco variety. Assign the weights with the deviation maximization, and comprehensive evaluate the tobacco varieties with choosing the mean area and the mean circumference as the evaluation vector. This method reflects the high recognition, merits and coordination of index.6. Validate the methods with the test data of tobacco varieties in North China variety regional test in 2007 and 2008. The result with deviation maximization was the same with the coefficient of variation, and the evaluation results with this weighting method were in accordance with the actual performance.So this is a btter weighting method in omprehensive evaluation of tobacco varitety. The weighting method with TOPSIS had a different results with the coefficient of variation, and the evaluation results differ from the practical performance. So this method need further improvement . The comprehensive evaluation of tobacco variety in this article had a consistent results with the GRA and actual performance. So this method can be a direct and visual comprehensive evaluation on tobacco variety.