基于近红外技术的主要进口国棉纤维快速识别  

Rapid identification of main imported national cotton fibers based on near-infrared technology

作　　者：耿向阳 郑丽莎 赵涛[2] 周成凤[2] 刘俊 GENG Xiangyang;ZHENG Lisha;ZHAO Tao;ZHOU Chengfeng;LIU Jun(Qingdao Customs Technology Center,Qingdao,266002,China;Qingdao University,Qingdao,266071,China;Chengdu Customs Technology Center,Chengdu,610095,China)

机构地区：[1]青岛海关技术中心,山东青岛266002 [2]青岛大学,山东青岛266071 [3]成都海关技术中心,四川成都610095

出　　处：《棉纺织技术》2024年第10期79-82,共4页Cotton Textile Technology

基　　金：国家自然科学基金(52303058);新疆维吾尔自治区重大科技专项项目(2020A03002-2);生物多糖纤维成形与生态纺织国家重点实验室自主课题(探索课题)(2020ZKT21)。

摘　　要：为了满足市场和海关检验部门对棉纤维来源进行快速鉴别的要求,基于近红外光谱技术,通过定性定量建模方法建立快速识别棉纤维国别的预测模型。使用傅里叶变换近红外光谱仪采集了澳大利亚、巴西、布基纳法索和印度4个国家棉花的近红外光谱数据,进一步使用簇类独立软模式法(SIMCA)和偏最小二乘法(PLS)等化学计量学方法对不同国家的棉花样品进行了分类模型构建与优化。结果表明,使用SIMCA法可基本实现不同国家的棉花识别工作,识别准确率为94%;进一步通过赋值法结合PLS建模可以实现不同国家棉花的快速准确识别,模型决定系数R2为0.9880,外部验证成功率达100%。In order to meet the requirement of rapid identification of cotton fiber in the market and customs inspection department,based on near-infrared spectroscopy technology,a prediction model for quickly identifying cotton fiber countries through qualitative and quantitative modeling methods was established.The Fourier transform near-infrared spectrometer was used to collect near-infrared spectral data of cotton from Australia,Brazil,Burkina Faso and India.Further,chemometrics methods such as soft independent modeling of class analogy(SIMCA)and partial least squares analysis(PLS)were used to construct and optimize classification models for cotton samples from different countries.The results showed that cotton recognition work in different countries can basically be achieved using SIMCA method,the recognition accuracy was 94%.Further combining the assignment method with PLS modeling can achieve rapid and accurate identification of cotton from different countries,with a model determination coefficient of R2=0.9880,and an external validation success rate of 100%.

关 键 词：棉纤维 近红外 化学计量方法 模型 快速鉴别方法 

分 类 号：TS107[轻工技术与工程—纺织工程]