Determination of total protein and wet gluten in wheat flour by Fourier transform infrared photoacoustic spectroscopy with multivariate analysis

This work investigated using Fourier transform mid-infrared photoacoustic spectroscopy (FTIR-PAS) for determining of total protein and wet gluten contents in wheat flour. Standard methods and near infrared spectroscopy for determining the total protein and wet gluten as comparing results. FTIR-PAS spectral and reference data were analysed by principal component analysis. Multivariate calibration models were established to evaluate the predict performance for total protein and wet gluten. In whole wavenumber dataset, satisfied results were obtained with root mean square error of prediction (RMSEP) of 0.362 % and 0.229 %, with coefficients of determination (R2) of 0.84 and 0.96 for total protein and wet gluten, respectively. Optimal wavenumbers with 23 and 31 were selected by successive projections algorithm (SPA) on the premise of reducing operation time and dimension data. Reduced models showed limited but acceptable RMSEP of 0.443 % and 0.315 %, R2 of 0.75 and 0.90 for total protein and wet gluten, respectively. Coefficients of determination in external validation were higher than 0.82 for total protein and wet gluten models. Comparing for performance of models by near infrared spectroscopy highlighted feasible FTIR-PAS spectral datasets can be predicted the content of protein in wheat flour.