Research on Content Measurement of Textile Mixture by NearInfrared Spectroscopy based on Principal Component Regression

摘要

A new method for accurate measurement of content of textile mixture by use of Fourier transform near infrared spectroscopy is put forward. The near infrared spectra of 56 samples with different cotton and polyester contents were obtained, in which 41 samples, 10 samples and 5 samples were used for the calibration set, validation set and prediction set respectively. Principal component analysis (PCA) was utilized for the spectra data compression. Principal component regression (PCR) model was developed. It indicates that the MAE is within 2.9% and the RMSE is less than 3.6% for the validation samples, which is suitable for the prediction of unknown samples. The PCR model was applied to predict unknown samples. Experimental results show that this approach by use of Fourier transform Near Infrared Spectroscopy can be used to quantitative analysis for textile fiber.