Analysis of pork adulteration in minced mutton using electronic nose of metal oxide sensors

The aims were to detect the adulteration of mutton by applying traditional methods (pH and color evaluation) and the E-nose, to build a model for prediction of the content of pork in minced mutton. An E-nose of metal oxide sensors was used for the collection of volatiles presented in the samples. Feature extraction methods, Principle component analysis (PCA), loading analysis and Stepwise linear discriminant analysis (step-LDA) were employed to optimize the data matrix. The results were evaluated by discriminant analysis methods, finding that step-LDA was the most effective method. Then Canonical discriminant analysis (CDA) was used as pattern recognition techniques for the authentication of meat. Partial least square analysis (PLS), Multiple Linear Regression (MLR) and Back propagation neural network (BPNN) were used to build a predictive model for the pork content in minced mutton. The model built by BPNN could predict the adulteration more precisely than PLS and MLR do.