Abstract With sustainable concept developing, the recycled nylon in the textile industry is becoming more and more widely used. Because the physical and chemical properties of virgin nylon and recycled nylon are very close to each other in terms of appearance and morphology, crystallinity, etc., traditional detection methods are unable to effectively distinguish between them. In this study, oligomers were collected from both virgin and recycled nylon samples through high-performance liquid chromatography (HPLC) technology. Then, the classification was achieved through feature extraction and pattern recognition of the chromatographic data based on a ResNet model, which can achieve a recognition accuracy of 90.6 %. At the same time, the Permutaion Importance method was applied to interpret and visualize the decision-making process of the ResNet. It was shown that the distribution and content of caprolactam peaks made the most significant contribution to the identification between virgin and recycled nylon and so they could be used to train the model. The development of a rapid identification system can serve the textile quality supervision and green certification, and has important practical value for regulating the recycled fiber market and promoting the development of circular economy in textile industry.
Wei et al. (Thu,) studied this question.