Prediction of the ash content of wheat flours using spectral and chemometric methods

Abstract

Fourier transform infrared (FTIR) spectroscopy has been considered an important
method in evaluating structural properties of biomolecules. The assignment of the
spectra indicated combination bands of the chemical bonds (O–H, C–H, C–C, etc.)
that compose the chemical constituents of biomass. This paper presents, for
several regional common wheat varieties (Triticum aestivum L.): Crina, Dropia,
Flamura, Gruia, Haiduc and wheat from Vaslui area, the comparative analyses of
the infrared absorption spectra for two spectral ranges: near-infrared (NIR) region,
in the corresponding wavelength of 700-1000 nm, and middle-infrared (MIR)
region, in the corresponding wavenumber 600-1600 cm-1. The positions of the
characteristic absorption peaks were observed at 877.49 and 941.03 nm in NIR
region and 756.50, 857.86, 929.75, 999.83, 1076.34, 1150.08, 1244.92, 1336.95,
1415.46 and 1538.35 cm-1 in MIR region. The information obtained from the
analysis of NIR-MIR spectra is used to rapidly determine some of physicochemical
parameters of the wheat species-specific part. The aim of this study is to
apply FT-NIR and FT-MIR spectroscopy to correlate those signals by the StatCorr
analyze. A fast technique such as chemometric analysis on the basis of NIR and
MIR spectra was used to predict the amount of ash from the flour sample. R
square of prediction model is 0.7 and Root Mean Square Error of Prediction
(RMSEP) using two Principal Component PCs is 1.5 % which certifies the
calibration model.