论文部分内容阅读
利用便携式近红外光谱仪对376个涤/棉混纺织物进行研究,利用定量分析模型中的偏最小二乘法(partical least squares,PLS)作为校正方法,结合涤/棉混纺织物中涤、棉含量设定的定性鉴别系数,建立了涤/棉混纺织物的半定量-定性分析校正模型。该模型对涤/棉混纺织物进行定性鉴别的同时得出其相对含量,分析结果具有半定量性质。在建模过程中,采用Savitzky-Golay导数法,消除噪声和基线漂移对光谱的影响,并研究了波段选择和不同预处理方法对定性校正模型的影响。纯棉的主要吸收峰位于1 400~1 600nm,纯涤的主要吸收峰位于1 600~1 800nm,随着涤或棉含量的增加,其相应的吸收峰强度增强,因此,建模波段以涤、棉主要吸收峰区间为基本波段,进行双向扩展,得到最佳波长区间1 100~2 500nm(相关系数0.6,波点数934)。利用所建校正模型对验证集样本进行预测,结果表明,在1 100~2 500nm处,预处理方法为Savitzky-Golay导数、多元散射校正与均值中心化相结合时,该模型评价参数较佳,其中RC(校正集相关系数)0.978,RP(验证集相关系数)0.940,SEC(校正标准差)1.264,SEP(预测标准差)1.590,样品预测正确率达93.4%。表明该定性分析校正模型能够较好地对涤/棉混纺织物进行半定量-定性预测。
A total of 376 polyester / cotton blend fabrics were studied by portable near infrared spectroscopy (NIRS). Partial least squares (PLS) in the quantitative analysis model was used as a calibration method. Combined with the polyester / cotton blend , The semi-quantitative-qualitative analysis calibration model of polyester / cotton blended fabric was established. The model of polyester / cotton blended fabric qualitative identification of the same time come to the relative content of the analysis results with semi-quantitative properties. In the modeling process, the Savitzky-Golay derivative method was used to eliminate the influence of noise and baseline drift on the spectrum. The effects of band selection and different pretreatment methods on the qualitative calibration model were also studied. The main absorption peak of cotton lies in the range of 1 400 ~ 1 600 nm. The main absorption peak of pure polyester lies in the range of 1 600 ~ 1 800 nm. With the increase of the content of polyester or cotton, the corresponding absorption peak intensity increases. Therefore, , The main absorption peak of cotton as the basic band, the two-way expansion, the best wavelength range 1 100 ~ 2 500nm (correlation coefficient of 0.6, the number of points 934). The proposed calibration model is used to predict the validation set samples. The results show that the pretreatment method is Savitzky-Golay derivative at 1 100 ~ 2 500 nm. When multivariate scatter correction and mean-centering are combined, the model has better evaluation parameters, The correlation coefficients of RC (correlation coefficient) 0.978, RP (correlation coefficient of validation set) 0.940, SEC (standard deviation of correction) 1.264 and SEP (prediction standard deviation 1.590) were obtained. The accuracy of prediction was 93.4%. It shows that the calibration model of the qualitative analysis can predict semi-quantitatively-qualitatively the polyester / cotton blended fabric.