Research Of Adulterated Goat Milk And Goat Milk Powder Rapid Identification By Electronic Nose

In recent years, domestic dairy industry has developed rapidly, and the market competition is becoming fierce, thus numerous dairy enterprises are striving for high quality raw milk. However, some raw goat milk and its products are adulterated, since unscrupulous suppliers could benefit from the short-term profits. For instance, using raw milk or expired reconstituted milk to partly substitute raw goat milk, or adding milk powder to goat milk powder. As a result, it will damage the products qualities and profits of dairy enterprises, and impair consumers’ physical health.The research chose electronic nose as fast and destructive detecting instrument, adding raw milk, expired reconstituted milk and milk powder to raw goat milk and goat milk powder, adulterants in different ratios were designed, using Loadings, principal component analysis(PCA) and linear discriminant analysis(LDA) to conduct qualitative judgement, using linear regression fitting to conduct quantitative analysis, then using DUPLEX method to divide training set and validation set, at last to recognize model by using Fisher Discriminant Analysis, multilayer perceptron neural network etc.Through the response diagram of raw goat milk and goat milk powder adulterated with three others exotic substances by using electronic nose we can see that: Different exotic substances had different atlas, and along with the increased ratio of addition, the variation differences of response signals expanded. Therefore different milk samples can be distinguished through response difference generated by electronic nose.The electronic nose could distinguish the control group’s samples from the samples contained three others exotic substances during qualitative analysis. The distinguish results of raw goat milk adulterated with raw milk were poor, and they were partly overlapping; yet to raw goat milk adulterated with expired reconstituted milk and goat milk powder adulterated with milk powder, both PCA and LDA could distinct samples in different adulterate ratio, and the distinction effect of LDA was better than PCA.To conduct the quantitative analysis, a linear regression fitting model was established. All the determination coefficients were above 88.4%, which means the model has generalization ability, the prediction accuracy reached 0.1%. Therefore the method of using electronic nose to qualitatively and quantitatively detect adulterants in raw goat milk and goat milk powder rapidly is feasible.During the process of model establishment, using Fisher Discriminant Analysis and multilayer perceptron neural network to recognize the model, the accuracy of training set and validation set of FDA reached 94% and 87% respectively, and the accuracy of training set and validation set of MLPN also above 100% and 85%, all of them had high prediction accuracies, thus good effect of prediction was obtained. So both FDA and MLPN can be served as effective prediction tools to help electronic nose to detect exogenetic dairy that added to raw goat milk and goat milk powder, and the prediction effect of MLPN was better than FDA.