高等学校化学学报 ›› 1999, Vol. 20 ›› Issue (S1): 27.

• Atomic Spectrometry • 上一篇    下一篇

Comparison of Neural Networks and Theoretical Correction Models

LUO Li-Qiang1, WU Xiao-Jun1, GAN Lu1, MA Guang-Zu1, YE Qingpei2   

  1. 1. National Research Center of Geoanalysis, Beijing, 100037, P. R. China;
    2. National Geological Museum of China
  • 出版日期:1999-12-31 发布日期:1999-12-31

Comparison of Neural Networks and Theoretical Correction Models

LUO Li-Qiang1, WU Xiao-Jun1, GAN Lu1, MA Guang-Zu1, YE Qingpei2   

  1. 1. National Research Center of Geoanalysis, Beijing, 100037, P. R. China;
    2. National Geological Museum of China
  • Online:1999-12-31 Published:1999-12-31

摘要:

In x-ray fluorescence spectrometry, matrix effects have to be corrected. The empiric equations, the theoretical coefficients and the fundamental parameters can be applied in the purpose. The classical matrix correction models include mainly Lachance-Trail,Rousseau, De Jongh, Claisse-Quentin, COLA algorithms,etc. The good results can be obtained by using a number of standards or the theoretical Alpha coefficients if the range of component concentrations in samples is not very wide. When unknown samples with a wide range of concentrations are predicted, a recalculation process of the theoretical Alpha coefficients has to be made. In addition, a spline function may be chosen to split wide range in several segments.

Abstract:

In x-ray fluorescence spectrometry, matrix effects have to be corrected. The empiric equations, the theoretical coefficients and the fundamental parameters can be applied in the purpose. The classical matrix correction models include mainly Lachance-Trail,Rousseau, De Jongh, Claisse-Quentin, COLA algorithms,etc. The good results can be obtained by using a number of standards or the theoretical Alpha coefficients if the range of component concentrations in samples is not very wide. When unknown samples with a wide range of concentrations are predicted, a recalculation process of the theoretical Alpha coefficients has to be made. In addition, a spline function may be chosen to split wide range in several segments.

TrendMD: