Studies On Discrimination Of Amino Resin,Oxytetracycline And Its Hydrochloride Adulteration In Dairy Cow Feed By Near Infrared Spectroscopy

In recent years, a series of harming people and livestock health and food safety events have happened because of lack of understanding and improper use for feed additives, and regarding the poisonous and harmful material as feed additives, such as the melamine event, antibiotics abuse problem and so on. According to the problems and to meet the current testing needs of rapid, nondestructive, this study was aimed at exploring the feasibility and methods of qualitative identification and quantitative prediction amino resin, oxytetracycline and its hydrochloride as adulterant in milk cow feed concentrate supplement by NIRS in order to provide suggestions for detection of amino resin and antibiotics.The followings are the experiment results:the full spectrum was used and the tests were conducted using the feed through20meshes. For qualitative identification models, appropriate pretreatment chosen, compared the results of PLS-LDA, PCA-BPNN and PCA-SVM, and then the optimal result was the PCA-SVM model. The optimal prediction accuracy rates of MF and UF were100%.Good quantitative models of amino resin were established by the classical PLS method based on choosing suitable pretreatment methods. For MF, calibration set were selected by SPXY, the RMSEP and R of testing set were0.0010and0.9936, respectively. For UF, the RMSEP and R of testing set were0.0048and0.9913, respectively.Conduct qualitative detection of oxytetracycline and its hydrochloride. The optimum was that CARS was used to choose variables, and then models were established by PLS-LDA. The results were as follows:for oxytetracycline, the number of variables was80, and the model misclassification rate and the model predicted error rate were0.0900and0.0417, respectively. For oxytetracycline hydrochloride, fewer variables66were selected, and the model misclassification rate and the model predicted error rate were0.0104and0.1250, respectively.In conclusion, amino resin not less than5%of adulteration in the milk cow can be identified by NIRS. Quantitative models of amino resin established by PLS can well predict the amino resin concentration0.5%-2.0%. Yet for the qualitative discrimination of oxytetracycline and its hydrochloride, severe spectrum collinearity was considered and useful variables were selected by CARS which combining with PLS-LDA would make the data processing speed faster and ensure the accuracy of recognition results. These results prove that qualitative identification and quantitative detection amino resin is feasible by NIRS, so qualitative discrimination oxytetracycline and its hydrochloride is.