The classification of Iranian wheat flour varieties using FT-MIR spectroscopy and chemometrics methods

Wheat varieties are grown in many countries throughout the world to produce grain and flour for foods, an outstanding and remarkable foodstuff that fulfills human nutritional needs. Methodologically, there is little known about how to distinguish accurately the type of wheat flour and, in the meantime, to restrain food fraud. Therefore, non-destructive methods are necessary for identification, in turn, the spectroscopy method has the potential to be one of the most efficient non-destructive methods. Given the importance of high-purity wheat flour for consumers and food industries, this research introduces the five most cultivated flour varieties in Iran that were classified by FT-MIR spectroscopy. Different preprocessing techniques and unsupervised and supervised models investigated the spectral data of flour varieties. The best outcome came with the standard normal variate (D2 + SNV) preprocessing method and with all supervised models. The accuracy of Linear Discriminant Analysis (LDA), Support Vector Machine (SVM), and Artificial Neural Network (ANN) was 100% for the test dataset. In conclusion, it is recommended flour industries use the FT-MIR spectroscopy and chemometrics techniques, the combined form of these techniques helps these industries be aware of the purity of produced flour as possible.