Study On Discrimination System Of Distilled Spirits Based On Three-dim Fluorescent Spectra Technique

As one kind of well-known spirits, Chinese distilled spirits have been blemished by fake and shoddy products. Several years ago, the fake Fenjiu had made many consumers loss of sight and even die. In recent years, although fake spirits made by industrial alcohol have disappeared, the ones made by edible alcohol are very common. People are paying more and more attentions to the problem that how to discriminate the Chinese spirits. Now, there are two ways to distinguish Chinese spirits. The first one is by degusting, and the other one is by the method of chromatographic analysis. The former is time wasted and costly. As a result, it can't be used far and wide. So, find a simple and convenient method of distilled spirits discrimination is very helpful for extinguishing the fake and shoddy products.There is a long history and a rich culture of distilled spirits in China. Over the years, we have been trying to analyzing and discriminating the distilled spirits by photoelectric equipments. But the composition of them is too complex, at present we mainly discriminate spirits by tasting. Our testing technology of spirits has made much progress, but we are also behindhand than western countries, especially in the respects of methods and equipments. As the reason of different materials and techniques, distilled spirits have special flavors. We can distinguish distilled spirits by the method of spectroscopic analysis.From large numbers of researches, we have found that different kinds of distilled spirits have almost the same kinds, but different quantities of substances. Obviously, to quantificationally analyze hundreds of monomers is not realistic and necessary. Spectral analysis has the strongpoint of integrity and non-destruction. Now it has been using in many respects. Over the years, technology of spectrum has been used in the respect of food discrimination. But most of them are using two-dimensional infrared spectrum, and their precision is not so desirable.This study has 95 kinds of distilled spirits (with 10 brands) were selected as the objects and their absorption and three-dimensional fluorescence spectra induced by ultraviolet light (200nm-500nm) have been studied respectively. We have extracted the characteristic parameters of their three-dimensional fluorescence spectra and do a research about them. As a result, we have found that it is effective to discriminate different distilled spirits by three typical parameters, namely the number of the major summit, position of the major summit and optimum wavelength. In order to prove that it is efficient to discriminate distilled spirits by the three parameters, cluster analysis as a kind of statistical method was used in this study. The fact has proved that the classification's accuracy rate of distilled spirit with different brands is very high. In the meanwhile, many modeling methods such as LS_SVM, SVM, PNN had been used to manipulate the data. Finally, LS_SVM was proved to be the best method. It can get the most precise outcome.All the data was stored into a database which was built with SQL Sever. On the basis, we have developed a system by VB to discriminate different distilled spirits. Its strongpoint is that it has high accuracy rate and is convenient to be used in the respects of industry manufacture and commodity circulation. It is significant to fight against fake products and protect the healthy security.