论文部分内容阅读
在93％-97．4％浓度范围内，利用维生素E（VE）在6061－5246cm－1处的近红外吸收峰峰面积积分值和其浓度关系建立的回归方程为：Y=103．43－0.078624X。用此回归方程对已知浓度的样品进行预测，误差及相对误差均在-0．79％－0．9％内。在较宽浓度范围80％-97％之间的VE，用PLS算法选择不同的数据预处理方法，对近红外吸光度值和其含量的关系建模，并用已知浓度的VE进行校验。结果表明，选择矢量归一的预处理方法，能够达到很好的预测效果。利用近红外光谱法测定VE的含量，并选择合适的算法对数据进行处理，具有操作简便、准确和快速等优点。
The regression equation established by the relationship between the concentration of vitamin E (VE) at 6061-5246 cm-1 and its concentration in the range of 93% -97.4% was: Y = 103.43- 0.078624X. Using this regression equation to predict the samples with known concentration, the errors and relative errors are within -0.79% -0.9%. In a wide range of concentrations of VE between 80% and 97%, different data preprocessing methods were selected using the PLS algorithm, the relationship between NIR absorbance values and their contents was modeled and verified with VE of known concentration. The results show that the selection vector normalized pretreatment method can achieve good prediction results. The use of near infrared spectroscopy to determine the content of VE, and select the appropriate algorithm for data processing, with the advantages of easy, accurate and rapid operation.