Integration of near-infrared spectroscopy and comparative principal component analysis for flour adulteration identification

Abstract

Flour, as a critical component of the dietary structure, its quality and safety assurance is of great significance. The combination of near-infrared (NIR) spectroscopy and chemometrics was proposed to identify the adulterated flour in three different brands. This study obtained the adulterated samples with different concentrations of talcum powder, and measured 20 spectral data corresponding to each concentration. Comparative Principal Component Analysis (cPCA) has a constraint effect on the background dataset and can reduce the interference of background factors. The results showed that the cPCA algorithm successfully eliminated brand-related factors when identifying adulterated flour, and achieved adulterated discrimination with a concentration as low as 0.3%. This study presents a practical approach for identifying flour adulteration, effectively tackling the challenge of background factors on feature extraction in data dimensionality reduction models. By addressing this issue, it paves the way for developing more accurate and reliable adulteration detection models.