The Model Establishment And Preliminary Study On The Mechanism Of Mutton Origin Traceability By Near Infrared Spectroscopy

Origin traceability is an important part of food traceability, it can protect food origin and ensure food safety. In recent years, the outbreaks of zoonosis such as sheep brucellosis and scrapie have threaten food safety and people health severely. It is important to develop a quick and accurate technology of mutton origin traceability for market management and ensuring food safety. The technology of mutton origin traceability by near infrared spectroscopy (NIRS) coupled with pattern recognition methods was studied in the paper.The aim of the research was to establish the mutton origin traceability model and predict the mechanism of mutton origin traced by NIRS. Specific results were as followes:(1) The mutton origin traceability model by NIRS coupled with SIMCA was developed. The results showed that the model of mutton origin traceability can be built up very well when original NIR spectra was preprocessed by five point Smooth and Multiplicative Scatter Correction (MSC) in the spectra region between 11995 cm-1 and 3999 cm-1 and the optimal number of principal component was 5, 6, 8, 7, 5 for Shandong, Hebei, Neimeng, Ningxia, Xinjiang. The recognition rate of calibration set for five origin models was 95%, 100%, 100%, 100%, 100% and the rejection rate of calibration set was 99%, 100%, 99%, 100%, 100% respectively. The recognition rate of validation set was 100%, 83%, 94%, 81%, 88% respectively and the rejection rate of validation set were all 100%.(2) The mutton origin traceability model by NIRS coupled with SVM was developed. Using the radial basis function (RBF) as the kernel function, the best model was built up when original NIR spectra was preprocessed by MSC, and penalty coefficient of C was 100000 and width parameter ofγwas 0.01. Under this condition, the classrate of calibration and validation set of mutton from Shandong, Hebei, Neimeng, Ningxia, Xinjiang region model were all 100%.(3) The mutton origin traceability model by NIRS coupled with ANN was build up. When original NIR spectra preprocessed by MSC, and the spectral data compressed by PCA was regarded as input vector, and the nodes of input layer, pattern lyer and output layer was 10, 8, 5, respectively, the model was built up well and the classrate of calibration and validation set of mutton from Shandong, Hebei, Neimeng, Ningxia, Xinjiang region were all 100%.(4) Comparison of the results of SIMCA, SVM and ANN models, it showed that SVM and ANN models were better than SIMCA model. The classrate of SVM and ANN models were all 100%.(5) Using Principle Componeng analysis(PCA),the Loadings analysis and Variance analysis, the relationship between NIR spectra and nutrient such as water, fat and protein of different geographical origins was found. It was predicted that due to the difference of environment and feeding conditions of mutton from different regions, the content and structure of water, fat and protein were different.The NIRS could reflect the difference, and the diversity could be picked up by PCA and the mutton origin traceability model could be built up coupled with pattern recognition methods.The results of this study indicated that NIRS coupled with pattern recognition methods was a feasible way for mutton origin traceability.