Study On Highly Efficient Nir Recognition Technology For Waste Polyester Fiber Products

The rapid identification and sorting of polyester fabrics were a major problem in the recovery of waste polyester fabrics.However,the traditional method was time-consuming,laborious and low recognition rate for the identification of textile fabrics.A total of 493 samples,including pure polyester textile,pure cotton textile,pure nylon textile,pure wool textile,polyester-cotton textile,polyester-nylon textile and polyester-wool textile,were studied in this article.The content of the component was measured by the national standard method before the original near infrared spectrum of the sample was tested by a portable near infrared instrument.The qualitative analysis and quantitative analysis model of common fabric was established.The 269 samples were used in calibration set while the remaining 224 as a validation sample.A rapid identification analysis model of common fiber products was established by the Python programming language with principal component analysis and support vector machine classification algorithm.For verify the accuracy and practicability of the model,224 samples were chosen as an validation set,the correct recognition rate of polyester fabric reached up to97.4%.It showed that the qualitative analysis model can be used to identify pure polyester fabric from the waste textiles.In the range of 1000 ～ 2500 nm band,the original NIRS was pretreated by preprocessing method,including means of mean centralization,differential first derivative,S-G smoothing and OSC.The near infrared quantitative analysis model of waste polyester fiber products was established by partial least squares which as the correction method and the main factor of the model was 7.The correlation coefficient of calibration and the correlation coefficient of prediction was 0.994 and 0.989,the Standard Errors of Calibration and the Standard Errors of Prediction were obtained to be 1.832 and 2.065.For verify the accuracy and practicability of the model,224 samples were chosen as an validation set,the correct recognition rate of polyester fabric reached up to 96.2%.It shows that the model can be used to distinguish pure polyester fabric from waste textiles.