Application of 2D Correlation Infrared Spectroscopy to Identification of Adulterated Milk Using NPLS-DA Method

In this paper, a new method for discriminate adulterated milk was proposed by combining two-dimensional (2D) correlation infrared spectroscopy with multi-way partial least squares discriminate analysis (NPLS-DA). 64 pure milk samples were collected and four adulteration types of milk with urea, melamine, tetracycline and glucose were prepared, respectively. IR spectra of pure milk and adulterated milk samples in the region of 900-1700cm-1 were measured. Based on the characteristics of 2D correlation infrared spectra, the discriminant model of pure milk and adulterated milk was conducted by using NPLS-DA method. The prediction rate of unknown samples was 90.7%. Finally, compared with NPLSDA and partial least squares discriminant analysis (PLS-DA) models, the NPLS-DA model showed excellent performance for prediction. This study demonstrates that 2D correlation spectra coupled with NPLS-DA is a promising way for rapid and non-expensive discrimination of adulterated milk.