高级检索

    复烤片烟常规化学成分的傅里叶变换近红外光谱法的模型转移

    Model Transfer of Routine Chemical Components in Redried Lamina on Fourier Transform Near Infrared Spectroscopy

    • 摘要: 为实现复烤片烟常规化学成分的模型在不同品牌傅里叶变换近红外仪器上的使用与共享,以贵州产区复烤片烟样品为研究对象,利用Kennard-Stone算法选择标准样品,将偏移量校正(BC)、截距斜率校正(SBC)和光谱空间转换(SST)等3种模型转移算法应用于不同品牌傅里叶变换近红外仪器的模型转移,并对3种模型转移算法的转移结果进行分析。结果表明:将复烤片烟常规化学成分的主机模型直接应用于从机预测时,主机和从机的预测值之间存在显著性差异;采用BC、SBC和SST可以实现不同品牌傅里叶变换近红外仪器的模型转移,其中SST转移结果最优。

       

      Abstract: In order to realize the use and share of model of routine chemical components in redried lamina on Fourier transform near infrared (NIR) instruments of different brands, the samples of redried lamina produced in Guizhou area were taken as the research object. The standard samples were selected by Kennard-Stone algorithm. Three model transfer algorithms, including the bias correction (BC), slope bias correction (SBC) and spectral space transformation (SST), were applied to model transfer on Fourier transform NIR instruments of different brands, and the transfer effects of the three model transfer algorithms were analyzed. The results showed that when the model of routine chemical components in redried lamina built from master instrument was directly applied to the slave instruments, there was significant difference in the predicted values between these instruments. The model transfer of Fourier transform NIR instruments of different brands could be realized by BC, SBC and SST. Among them, SST gave the best transfer results.

       

    /

    返回文章
    返回