基于近红外光谱和深度学习算法的羊毛混纺废旧纺织品定性分析  

Qualitative analysis of waste wool based on near infrared spectrum and deep learning algorithm

作　　者：邱迅 刘媛媛 郭晟材 刘冬志 王翀 刘俊 徐淑华 杨莉 龚龑 QIU Xun;LIU Yuanyuan;GUO Shengcai;LIU Dongzhi;WANG Chong;LIU Jun;XU Shuhua;YANG Li;GONG Yan(School of Textile and Clothing,Xinjiang University,Urumqi,Xinjiang 830017,China;School of Material Design and Engineering,Beijing Institute of Fashion Technology,Beijing 100029,China;Tianjin Jiuguang Technology Development Co.,Ltd.,Tianjin 300450,China;Xinjiang Huanjiang Lüyuan Environmental Protection Technology Co.,Ltd.,Urumqi,Xinjiang 830019,China;Urumqi Customs,Urumqi,Xinjiang 830011,China;Xinjiang Ruyi Textile and Garment Co.,Ltd.,Shihezi,Xinjiang 832000,China;Xinjiang Ruyi Fashion Clothing Co.,Ltd.,Shihezi,Xinjiang,832000,China;School of Mechanical and Electrical Engineering,Tarim University,Tarim,Xinjiang 843300,China)

机构地区：[1]新疆大学纺织与服装学院,新疆乌鲁木齐830017 [2]北京服装学院材料设计与工程学院,北京100029 [3]天津九光科技发展责任有限公司,天津300450 [4]新疆环疆绿源环保科技有限公司,新疆乌鲁木齐830019 [5]乌鲁木齐海关,新疆乌鲁木齐830011 [6]新疆如意纺织服装有限公司,新疆石河子832000 [7]新疆如意时尚服装有限公司,新疆石河子832000 [8]塔里木大学机械电气化工程学院,新疆塔里木843300

出　　处：《毛纺科技》2023年第1期91-98,共8页Wool Textile Journal

基　　金：新疆维吾尔自治区重大科技计划项目(2020A03002-4);新疆生产建设兵团科技攻关计划项目(2019AB014)。

摘　　要：从废旧纺织品中回收羊毛纤维并重新制成羊毛纱线或其他羊毛纺织品是一种经济又环保的废旧纺织品回收方式。为了实现废旧纺织品中羊毛纤维开环或闭环的机械回收,需要使用分拣设备和分类算法对不同羊毛纤维含量的废旧纺织品进行自动分拣。为了完成自动分拣,使用近红外光谱仪对不同混纺比例的羊毛废旧纺织品进行扫描,获得对应废旧纺织品近红外光谱的数据集。对数据集进行预处理后,采用深度学习算法对数据进行建模,并对获得的模型进行测试,以验证其准确率和效率,结果证明该模型在废旧纺织品的近红外光谱数据集上有良好的准确率。Recycling wool fibers from used textiles and remaking them into wool yarn or other wool textiles is an economical and environmentally friendly way to recycle used textiles.In order to realize the mechanical recovery of wool fibers in waste textiles with open or closed loops,it is necessary to use sorting equipment and sorting algorithms to automatically sort waste textiles with different wool fiber content.In order to complete the automatic sorting,a near-infrared spectrometer was used to scan wool waste textiles with different blended proportions to obtain a dataset corresponding to the near-infrared spectrum of waste textiles.After the dataset is preprocessed,the deep learning algorithm was used to model the data,and the obtained model was tested to verify its accuracy and efficiency,and the results prove that the model has good accuracy in the near-infrared spectral dataset of waste textiles.

关 键 词：近红外光谱 深度学习 卷积神经网络 废旧纺织品回收 羊毛回收 定性分析 

分 类 号：TS102.311[轻工技术与工程—纺织工程] O657.3[轻工技术与工程—纺织科学与工程]