No abstract

No abstract

Chinese national standards regulate that the melamine in infant food should not exceed 1 mg/kg. To quickly and accurately identify whether the melamine in raw milk exceeds the limit, near-infrared transmission (NIRT) spectra of 220 raw milk samples and 380 melamine-adulterated raw milk samples were collected over the spectral range of 900-1,700 nm. Different spectral preprocessing and dimension reduction methods were used to preprocess the data of full spectra with 425 wavelengths. The qualitative prediction models of partial least squares discriminate analysis (PLS-DA) and support vector machine (SVM) were built based on full spectra and reduced data.The results showed that the SVM was better than PLS-DA in identifying the melamine-adulterated raw milk whose melamine contents exceeded the limit of 1 mg/kg. The established SVM model using full spectra had the optimal prediction performance with the accuracy rate of 92.67%, sensitivity of 89.09%, and specificity of 94.74%. The study indicates that the raw milk whose melamine contents exceed the limit of 1 mg/kg could be identified by using NIRT spectroscopy. Practical ApplicationsThis study demonstrated that the raw milk whose melamine contents exceed the limit of 1 mg/kg could be identified by using near-infrared transmission spectroscopy.The results provide technical support and effective information for developing portable instrument to detect melamine-adulterated raw milk in situ or in field.