Research And Application Of XRF Spectral Simulation Algorithm Based On Fundamental Parameter Method

Taking X-ray fluorescence analysis as the discipline background,thesis is entitled to the general project of National Natural Science Foundation of China.It creatively puts forward the algorithm model of Fundamental Parameters（FP）to simulate accurate X-ray fluorescence spectrum（XRF）.The algorithm can quickly build an XRF element spectrum database with a large amount of data,which provides data support for X-ray fluorescence analysis.The overall structure of the article is as follows:（1）According to the Sherman formula,the article obtained the calculation relationship between the element content and the theoretical intensity of X-ray fluorescence.Carry out research on various fundamental parameters:Interpolate and fit the photon cross section to calculate the mass absorption coefficient.Dimension reduction analysis of X-ray excitation rate was carried out for L shell spectral line fraction.And propose the calculation formula for the excitation factor.Complete the fundamental parameters database of elements with atomic numbers 11～92,and developed the program code for FP method to simulate XRF spectrum on the MATLAB platform.（2）A method based on the Fundamental Parameters algorithm to quickly simulate and accurately generate the XRF spectrum of the material is realized.The main process is as follows:1.Input the fundamental parameters of the element into the Sherman formula to obtain the discrete X-ray fluorescence theoretical intensity spectrum of the sample.2.Research on Detector Response function and Gaussian energy broadening,convert discrete X-ray fluorescence intensity spectrum into continuous X-ray fluorescence spectrum;3.Use genetic algorithm to calculate and optimize instrument factor parameters,and finally obtain complete（K and L shell）and accurate XRF spectrum for any elemental material.（3）The XRF spectra of 72 standard alloy samples and 60 standard soil samples were simulated by Monte Carlo N Particle Transport Code（MCNP）,and the spectra simulated by the FP algorithm were compared with the simulation results of MCNP.The results show that the coincidence of the spectra is great,and the goodness of fit R~2 is greater than0.95.In addition,compared with the results measured by the X-ray fluorescence analyzer,the results show that the measured data of the sample spectrum and the FP method simulation are almost completely coincident（R~2>0.98）.Therefore,the spectra simulated by FP can partially replace the XRF data of the instrumental measurement of material materials.Furthermore,the spectra generated by FP have the ability to simulate heavy elements with higher spectral accuracy for simulated heavy elements.（4）Taking the quantitative analysis of precious metals Au,Ag,and Pt as an example,the FP method is used to simulate and generate a training set of spectral data based on a large number of samples,and a training set of experimental spectral data is established based on the existing experimental samples.The two datasets are extracted by principal components and input to the BP（Back Propagation）neural network.The verification results show that the R~2 of the FP prediction model is greater than 0.99,while the R~2 of the experimental prediction model is only 0.88.The spectral data simulated by FP method can solve the problem of insufficient sample data and improve the accuracy of element quantitative analysis in XRF spectral inversion.In summary,the XRF spectral model algorithm derived from the basic parameter method realizes the rapid and accurate simulation of spectral data,and the simulated spectra provide data support and theoretical basis for the quantitative analysis of elements in spectroscopy.