数据融合在现场X射线荧光分析中的应用  

Application of data fusion in-situ X-ray fluorescence analysis

作　　者：邹永祥[1] 吴建平[2] 

机构地区：[1]宜宾学院应用技术学院,宜宾644000 [2]成都理工大学核技术与自动化工程学院,成都610059

出　　处：《核技术》2009年第7期515-520,共6页Nuclear Techniques

基　　金：四川省应用基础项目(2006J13-018)资助

摘　　要：现场X射线荧光分析中,在样品测量面形状、湿度、密度和矿物颗粒度等差异对特征X射线照射量率影响的基础上,阐述了数据融合和模块化神经网络的基本思想,结合两者建立描述和校正上述干扰因素的神经网络融合模型、算法及学习方式。以该校正模型对试验数据计算和仿真,将测量面形状、湿度、颗粒度干扰造成的误差降低到小于10%;实验结果表明建立的校正模型有较好精度,基于模块化神经网络的数据融合技术能运用在现场荧光分析的干扰校正中。In this paper, effects of surface smoothness, moisture content, density and mineral granularity of the samples on X-ray fluorescence yield are analyzed, and principals of data fusion and modular neural network are introduced. A neural network data fusion model was established, with its learning algorithm for correcting the interfering factors. With the model, error caused by surface smoothness, moisture content and density could be reduced to less than 10%. The results prove that the mathematic models have high precision and the fusion technology of information can be used for in-situ correction of XRF analysis results.

关 键 词：现场X射线荧光技术 元素 干扰 数据融合 神经网络 

分 类 号：O657.62[理学—分析化学] TP391[理学—化学]