Discriminant analysis of milk adulteration based on near-infrared spectroscopy and pattern recognition

Abstract

Since the beginning of the 21st century, the issue of food safety is becoming a global concern. It is very important to develop a rapid, cost-effective, and widely available method for food adulteration detection. In this paper, near-infrared spectroscopy techniques and pattern recognition were applied to study the qualitative discriminant analysis method. The samples were prepared and adulterated with one of the three adulterants, urea, glucose and melamine with different concentrations. First, the spectral characteristics of milk and adulterant samples were analyzed. Then, pattern recognition methods were used for qualitative discriminant analysis of milk adulteration. Soft independent modeling of class analogy and partial least squares discriminant analysis (PLSDA) were used to construct discriminant models, respectively. Furthermore, the optimization method of the model was studied. The best spectral pretreatment methods and the optimal band were determined. In the optimal conditions, PLSDA models were constructed respectively for each type of adulterated sample sets (urea, melamine and glucose) and all the three types of adulterated sample sets. Results showed that, the discrimination accuracy of model achieved 93.2% in the classification of different adulterated and unadulterated milk samples. Thus, it can be concluded that near-infrared spectroscopy and PLSDA can be used to identify whether the milk has been adulterated or not and the type of adulterant used.