A Identification Method Study Of Pork,Beef And Mutton Based On VIS/NIR Spectroscopy

Xinjiang is the minority nationality inhabited areas, at present, there are 47 ethnic minorities, among them have 13 native ethnic groups, most of the people believe in Islam,they eat beef and mutton as their principal food. Xinjiang is a big province of animal husbandry in China,but the high price is not high quality mutton the phenomenon is prominent.The market is flooded with different and varieties different levels of mutton.Even adulteration incident occurred.How to identify meat is an urgent problem to be solved. Near infrared spectroscopy is one of the rapid development in recent years, high and new technical analysis, it has the characteristics of fast, accurate, non-destructive and so on,the detection has been widely used in the identification of food. This paper from the species, varieties, mixed meat three aspects, qualitative identification of meat.Using visible / near infrared spectroscopy scanning for the qualitative identification of meat,to establish prediction model.The main results are as follows:The identification of pork, beef and mutton samples,the results show that: linear discriminant analysis model is established by spectral characteristics, the recognition accuracy of pork, beef, mutton and prediction samples were 100%, 94%, 95.5%, the model is reliable. In addition, use of cluster analysis can be carried out better clustering of the pork sample, the accurate rate reached to 100%, the beef and lamb samples clustering result is not ideal.Collection 235 different varieties of mutton samples, extraction of the original visible / near infrared spectroscopy. Select three bands to carry on the principal component analysis,According to the principal component scores to establish the linear discriminant model. The results show that, when the number of principal components was 7, in the range of 400nm-430 nm, the calibration back substitution accuracy rate was 75.5%, the accuracy of cross validation was 73.4%, the validation set accuracy rate was 93.1%. The model established deal with the first derivative and SNV, w hen the number of principal components was23, 400nm-430 nm band discriminant accuracy rate reached 93.6%, the accuracy of cross validation was89.4%.Using the linear analysis to establish the pork, mutton mixed sample identification model. The classification model was significant, the accuracy of the calibration set and the cross validation accuracy rate reached 100%. Mixed samples of validation set there were 2 be mistaken, the correct recognition rate of the test set was 85%. Thus it can be seen, visible / near infrared spectroscopy technology can be used for the rapid identification of different species and different varieties of meat.