![](https://simg.instrument.com.cn/bbs/images/brow/em09502.gif)
原文我就不贴出来了,摘抄文章里的一段:
“For the evaluation of the resolution effect to
NIR measurements a demanding quantitative application, the so-called BTX experiment, was selected. The object is to quantify the components of mixtures of benzene(74-94%), toluene(0-20%), o-xylene(0-5%), m-xylene(0-5%)and p-xylene(0-5%). The study was base on 50 samples, carefully mixted from pure components.
For these samples, five sets of 150 spectra each were recorded at resolutions of 2cm-1, 4cm-1,8cm-1 and 16cm-1 with a Bruker Matrix FT-
NIR spectrometer. The number of scans was adaped to the resolution selected in such a way that the same measuring time was available for each record.
Of main interest are the standard errors of predictions for the xylene isomers, which represent a solid estimate for the precision limits for the quantification of liquid mixtures. The values are summarised in Table 1. PLS was applied for evaluating the optimum calibration. They are based on the spectra in the range between 4150cm-1 and 6300cm-1.
The results show almost
no relationship between resolution and the prediciton quality of calibration within the selected resolution range.This is due to the fact that the theoretical gain of information at higher resolution is balanced by the lower singal-to noise ratio. Therefore, 16cm-1 could be chosen without significant loss. ”
![]()
我就不一句一句翻译了,意思是说用2cm-1, 4cm-1,8cm-1 and 16cm-1不同分辨率进行对比,发现结果其实差不多,也就是说高分辨率是没用的