中文摘要：

为了实现纤维种类的快速鉴别，选用了棉、麻、毛、丝、天丝5种纤维，提出了一种用近红外光谱技术快速无损鉴别纤维品种的新方法。应用可见/近红外光谱漫反射技术测定各种纤维的光谱曲线，用主成分分析方法（PCA）对光谱数据进行模式特征分析，根据主成分的累积贡献率选用前6个主成分数进行建模和预测，通过建立最小二乘支持向量机模型（LS-SVM）对主成分分析模型进行优化，将前6个主成分作为最小二乘支持向量机的输入变量，建立PCA-LS-SVM模式识别模型，实现类别预测的同时也完成了数学建模与优化分析工作。5个品种的纤维训练集样本200个用于PCA-LS-SVM的模型的建立，对其余预测集样本50个进行验证,结果能准确的区分预测集的5种纤维。并提出主成分分析结合最小二乘支持向量机的光谱数据分析方法具有很好的分类和鉴别作用，为纤维品种快速鉴别提供了一种新方法，为维护消费者权益，保证纺织品质量，实现纺织原料及其制品的合理化生产与交易具有重要的意义。

英文摘要：

The aim of the present paper was to provide new insight into Vis/NIR spectroscopic analysis of textile fibers. In order to achieve rapid identification of the varieties of fibers, the authors selected 5 kinds of fibers of cotton, flax, wool, silk and ten- eel to do a study with Vis/NIR spectroscopy. Firstly, the spectra of each kind of fiber were scanned by spectrometer, and princi- pal component analysis （PCA） method was used to analyze the characteristics of the pattern of Vis/NIR spectra. Principal com- ponent scores scatter plot （PC1 × PC2 × PC3） of fiber indicated the classification effect of five varieties of fibers. The former 6 principal components （PCs） were selected according to the quantity and size of PCs. The PCA classification model was optimized by using the least-squares support vector machines （LS-SVM） method. The authors used the 6 PCs extracted by PCA as the in- puts of LS-SVM, and PCA-LS-SVM model was built to achieve varieties validation as well as mathematical model building and optimization analysis. Two hundred samples （40 samples for each variety of fibers） of five varieties of fibers were used for cali- bration of PCA-LS-SVM model, and the other 50 samples （10 samples for each variety of fibers） were used for validation. The