Fast Detection Of Fiber Content In Wool And Polyester Blended Fabric Based On Near- Infrared Spectroscopy

This paper studies the object textile fabrics and uses near infrared spectroscopy method and chemometric method, researching how to detect the fiber content of wool and polyester blended fabric. The main contents includes the following technical matters, the experimental detection of near infrared spectroscopy, samples collection, experiment parameter selection, spectrum preprocessing method selection, mathematical model building and so on.The experimental samples consist of three kinds, pure wool fabrics, polyester fabrics, and the mixture of wool and polyester fabrics. The former two kinds samples have different coloures and specifications.Among wool and polyester blended fabrics are own products of different mixture ratios from 0 to 90, and the other different mixture ratios wool and polyester fabrics.The instruments used in this study are Sup NIR-1500 series of portable analytical ones made by Focused Photonics Inc limited company located in Hangzhou. Wavelength coverage is from 1000 nm to 1800 nm.Detect the accuracy of the near-infrared spectroscopy through experiments. And then compare and analyze the experiments data, decide the conditions of experiments machines and experiments parameters like wavelength accuracy, distinguish ability, spectrum scan times. At the same time, we’ll analyze the affections on experiments results of environment humidity and temperature, samples fineness and colours.Collect the near infrared spectrum of samples, preprocess the spectrum and obtain the characteristic parameters. The preprocess methods include standardization, reciprocal, smoothness, signal correction. Through analysis and comparison, We select the mean value centralization and reciprocal and multiplicative scatter correction as the preprocess method. Then we can get the most valuable one from the mass information contained in the spectrum, eliminating the affection of of samples surface scattering on spectrum as well as system and random error. As a result, the stability and predictive ability of the model can be improved.we analyze the experiments result of principle component analysis, multiple linear regression model, least square method and artificial neural network. The data shows that the least square method has the best model results, which reports that wool content standard deviation is 0.0007 and the correlation coefficient of the predictive value and actual value is 0.99223. And the two parameters of Dacron content respectively are 0.0016 and 0.99161.We next revise and optimize the model, and predictive parse the samples. The absolute average value is less than 0.31%(wool :0.3057%,polyster fiber:0.3016%). And the standard deviation is less than 0.6%(wool :0.32%,polyster fiber:0.58%). The above results are within normal limits and shows the model can be used to detect the fiber content of mixture textile fiber. The results of this study shows the feasibility of the nearinfrared spectroscopy to detect the content of mixture fiber, and provide powerful evidence of near- infrared spectroscopy to detect the content of textiles.