Optimization Of Research And Application Of FP Method Analysis Technology Without Sample In EDXRF

Energy dispersive X-ray fluorescence analysis technology is one of the most advanced non-destructive assays and has wide ranging applications. This technology has advantages such as fast analysis, no sample destruction, wide detection range, reliable results and easy to implement online analysis and automatic analysis, etc.Since the German physicist Roentgen discovered x-rays in1895, research scholars worldwide has conducted in-depth research work in the field. Studies in X ray fluorescence analysis technology and application has passed more than one hundred years and achieved fruitful results. X-ray applications have penetrated into all aspects of the social economy industries and public life. Since the late1960s, many researchers in China carried out researches of energy dispersive X-ray fluorescence method actively. According to the fields the units involved in and experimental conditions, a large number of intensive researches from different aspects has conducted and made gratifying achievements. In1978, as the project, energy dispersive X ray fluorescence analysis method was discussed in the first activation analysis meeting of our county. After that, paper or report in this aspect were published and reported in each session of the national conference on activation analysis meeting and related professional meetings and a wealth of experience were accumulated. However, compared with foreign countries, our country still needs further work in terms of basic research and technology development.In X-ray fluorescence spectrum analysis, the relation between measured count rates with the test elements is generally nonlinear because of the matrix effect. Therefore, we need to correct matrix effect in order to analyze the samples accurately, it’s also an important subject which needs to be continued in the field of X ray fluorescence analysis. Traditional experience coefficient method heavily depends on a large number of standard samples similar to matrix, but the standard samples are misfit with test samples, at the same time, other production areas and research fields are also demand more for less sample and non-sample analysis.Based on the key problems above which exists in the X ray fluorescence analysis, relied on the National Natural Science Foundation of China" Research of EDXRF multi-element spectral library MC simulation and no sample analysis technology.(NSFC No.40974065)",carried out two aspects of research in fundamental parameters no sample analysis technology and the optimization of this method.The main research contents and results are as follows:(1) The theoretical research of fundamental parameters method without sample analysis.Through the study of the fundamental parameter method without the prototype analysis theory, mastered the fundamental parameter method without the prototype analysis technique, gained the computing method of key physical parameters in the method such as mass absorption coefficient, fluorescent yield, absorption jump factor and excitation factor, wrote a computing program according to fundamental parameter method without the prototype analysis(2) Application of fundamental parameter method without the prototype analysis technology in simple samples.Using fundamental parameter method without the prototype program to analysis two kinds of binary element samples (Fe-Cu and Ti-Fe), ternary element samples (Ti-V-Fe) and leaded brass alloy sample. In application of Fe-Cu binary system, the relative errors of Fe element are within7.4%, the relative errors of Ti element are less than3.9%; In Ti-Fe binary system, he relative errors of Fe element are within5.6%and the relative error of Cu element are within6.6%; there are good measurement effects in Ti-V-Fe ternary system and leaded brass samples, Analysis results show that, mixed sample measurements are matched with the given values and without prototype in analysis process, which verified the feasibility of fundamental parameter method without the prototype and the correctness of the physical parameters in this method, provides the support for later analysis of more complex samples.(3) Using BP network to carry out optimization research for fundamental parameter method without the prototype technology.Building the BP neural network on the MATLAB platform based on LM algorithm. This paper using relative intensity calculated via fundamental parameter method without the prototype program as input of the BP network, through the network training, predicting concentration of each element in the samples, then compare the predict concentrations with given values. The relative errors of Cu、Fe and Zn are lower than2.74%,1.74%and1.9%respectively, which proved that the BP neural network has good prediction effect. The combination of fundamental parameters method and neural network achieved the optimization of fundamental parameter method without the prototype technology.This subject is to carry out basic research work at the same time there is a degree of innovation in methods and technology. In the process of analyzing sample, it overcomes the absorption enhancement effects between elements in the complex samples finally, formed a set of optimized energy dispersive X fluorescence fundamental parameter method without the prototype measurement technology.There are two main innovation of this paper:(1) Through theoretical research, wrote the fundamental parameter method without the prototype calculation program, using the program to analysis the simple component system samples and series of lead brass alloy samples on the basis of experimental research, verified the feasibility of the fundamental parameter method without the prototype technology.(2) The Combination of conventional fundamental parameter method without the prototype and neural network method was applied to analysis samples, it overcomes the absorption enhancement effects between the elements, forms a set of optimized energy dispersive X fluorescence fundamental parameter method without the prototype technology system.This paper is from the basic theory of energy dispersive X fluorescence measurement, through the measurement of energy dispersive X fluorescence and gain the spectrums of different system samples prepared, in order to optimize the conventional fundamental parameter method without prototype, using the fundamental parameter method without prototype analysis program to calculate the relative intensity of each element which are as the input of neural network model, gain the predictive values and compare them with the given values, the experimental results show that this method can overcome the absorption enhancement effects between elements better, achieved the desired purpose.