Quantification of components of textile fabrics by using chemometric techniques with FT-NIR spectroscopic data

The study has attempted to develop chemometric modeling based method to quantify compositions of textile fabrics by FT-NIR spectroscopic data. Three calibration techniques such as: Principal Component Regression (PCR), Partial Least Square Regression (PLSR) and Artificial Neural Network (ANN) were assessed, and PLSR showed the best result. Several pretreatment techniques of spectral data have been evaluated, and Multiplicative Scatter Correction (MSC) performed the best. Results also shows that performance of PLSR was satisfactory for quantification of cotton (R2 ≈0.99), elastine (R2 ≈0.97) and polyester (R2≈0.94) when FT-NIR spectral data were pretreated with MSC. But for quantification of viscose in mixture fabric, efficiency of developed model was not upto the mark (R2≈0.75). Finally, the developed PLSR model with FT-NIR spectroscopic data pretreated with MSC could be used for quantification of cotton, elastine and polyester in textile fabrics rapidly and with comparatively low cost. Bangladesh J. Sci. Ind. Res. 57(4), 229-238, 2022