中文摘要：

将近红外光谱法和模型集群分析方法应用于毛涤混纺织物成分含量的快速无损测定。以近红外测量方法采集93个毛涤混纺织物的光谱信号,利用光谱预处理消除信号漂移的影响,在模型集群分析基础上,剔除异常样本,筛选出30个关键波长,采用偏最小二乘法（PLS）建立涤纶含量的预测模型。所建立模型的训练集相关系数r2为0.9827、交互验证均方误差（RMSECV）为3.26、预测均方根误差（RMSEP）为3.34,预测结果令人满意,适合于毛涤混纺织物中涤纶含量的快速、无损检测。

英文摘要：

Near-infrared spectroscopy as a rapid, non-destructively testing technique, has been widely used in the fiber product testing field. 93 polyester/wool samples were collected. Model population analysis method was employed to detect the outlier and select key variables after preprocessing the spectra by Savitsky-Golay derivative method. Partial least squares （PLS） calibration models were established by the optimal conditions to predict the content of polyester. Correlation coefficient of determination r^2, root-mean-square error of cross-validation （RMSECV） and root-mean-square error of prediction（RMSEP） were used to evaluate the quality of the model. The best models showed satisfactory predictions as measured by the r^2, RMSECV and RMSEP values： 0.982 7, 3.26 and 3.34. The prediction results were better than the whole spectra The results showed that the method was suitable for the fast and reliable determination of the content of polyester in polyester/wool product.