|
Keywords
|
Adulteration, FT-IR spectroscopy, Linear Discriminant Analysis (LDA), Preprocessing, Support Vector Machine (SVM), Wheat four
|
|
Abstract
|
Bread is a major source of human’s diet especially in developing countries and has significant role in human's healthy lifestyle. The addition of sodium dithionite (Na2S2O4) to flour for whitening has dangerous effects on human's health. In this research Fourier Transform-Infrared (FT-IR) spectroscopy with chemometric techniques is used to identify this compound in flour. After acquiring spectral data some preprocessing methods for removing uninformative effects and improving model results were applied. Principle Component Analysis (PCA) as unsupervised and Support Vector Machine (SVM), and Linear Discriminant Analysis (LDA) as supervised models were used. By comparing different preprocessing methods with models, The SVM model had the best classification result with 100% and 79.62% for two (pure and adulterated) and five (pure and adulterated with adulteration levels) class classification, respectively. This study showed the applicability of FT-MIR spectroscopy technology with chemometrics methods for detection of fraud in wheat flour.
|