【摘要】 在X射线荧光分析过程中,由于分析样品中元素的母质来源及形成条件不同,存在明显的吸收增强-基体效应。本文针对该类影响产生的偏差,通过研究归纳基体产生的特征和物理参数规律,提出合理正确的对样品进行分类是消除元素间基体效应的根本途径。通过神经网络方法建立样品自动分类模型,能够对各种不同的样品进行自动分类识别。经过初步应用,与传统方法进行对比,具有自学习、自组织的特点,不需要对样品做任何假设,所建立的自动分类模型稳定性好,可以多次运作,对未知样品的识别率较高,并且具有很强的容错、抗干扰能力,识别速度较快,基本上可以消除由于不同类样品元素间效应引起的基体效应影响。更多还原

【Abstract】 In XRF analysis process,the absorption intensification-matrix effect obviously exist,for the source and formation conditions of the analysed elements’ parent materials are different.This paper advances that to classify the samples reasonably and correctly is the basic way in view of the deviation produced by the influence and also by studying and concluding the character as well as physics parameter law produced by the matrix.The auto-classification model used in samples is set up by the method of NN.It can do auto-classifcation identification in different samples.Compared with the traditional way,it has the characteristic of self-study and self-organization,and needs no suppose to samples.The auto-classification model,with good stability and high identification rate to unknown samples and high rate to identification,can acts time after time and has the high ability of allowing fault and anti-jamming and can eliminate the matrix effect influence produced by the effect in different kind of sample elements.To promote the study that uses the method mentioned above in field and room EDXRF analysis technique can not only improve the application of EDXRF analysis technique in geonomy research and offer it with modern technique equip and advanced technology,but also be advantageous to spring up the geonomy new method and new technique.It has very important scientific significance and the applied future is abroad.更多还原