基于吸光度的波长筛选方法用于近红外光谱定量模型的优化

Optimization of near infrared spectroscopy quantitative model based on absorbance wavelength selection method

建立一种基于吸光度的波长筛选方法,以近红外光谱测定中成药制剂的多糖含量为例,对模型优化效果进行验证。考虑模型稳定性,在计算机平台上搭建一种新的样本集划分框架,基于吸光度筛选出最优波段为400~1882＆2072~2364 nm,建立偏最小二乘（PLS）模型得到的SEPAve、RP,Ave分别为27.13 mg L-1、0.856,与全扫描谱区（400~2498 nm）的PLS模型预测效果做比较。结果表明,基于吸光度的波长筛选方法,可以优选出高信噪比波长,从而提高了近红外光谱定量模型的性能。

To establish a screening method based on the absorbance wavelength,the polysaccharide content of the Chinese traditional medicine determination of near infrared spectroscopy as an example,to verify the effect of optimization model.Considering the stability of the model,the computer platform to build a new framework to optimize the sample set,absorbance wavelength of 400 ~ 1882 and 2072 ~ 2364 nm based on the established partial least squares（PLS） model of SEPAve,RP and Ave were 27.13 mg,0.856 L-1,and full scan spectral region（400 ~ 2498 nm） PLS model to predict the effect comparison.The results show that the wavelength of the high signal to noise ratio can be optimized by the wavelength selection method based on the absorbance,which can improve the performance of the quantitative model of the near infrared spectrum.