偏最小二乘法回归模型在分析毛涤混纺面料纤维含量中的应用

Using the partial least square regression method to analyze the fiber content of wool/polyester blended fabrics

以毛涤混纺面料为研究对象,利用近红外光谱法,结合相应的化学计量学方法建立不同的数学预测模型,进行毛涤混纺面料中纤维含量的分析研究。实验对比分析了主成分回归模型、多元线性回归模型、偏最小二乘回归分析模型的模拟结果。其中,偏最小二乘法回归分析模型得出毛纤维含量的校正集标准差(SEC)为0.000 7,校正集预测值与真实值相关系数(RC)为0.992 23,涤纤维含量的SEC为0.001 6、RC为0.991 61,为最优分析结果。研究结果为数学模型在混纺面料成分定量分析中的应用提供了有力的依据。

This paper studies the wool and polyester blended fabric,using the near infrared spectroscopy method and chemometrics method to build different mathematical models so as to detect the fiber content of wool and polyester blended fabric.We analyze the results of principle component analysis,multiple linear regression model,and partial least square method with the experiments.The partial least square method shows that it can obtain the best model result that the wool content standard deviation is 0.000 7 and the correlation coefficient of the predictive value and actual value is 0.992 23.And Dacron content and the relation coefficient are respectively 0.001 6 and 0.991 61.The results of this study provided a powerful evidence of using mathematical model to analyze fiber content of wool and polyester blended fabrics.