Research On Uranium Ore ? Spectrum Logging Data Analysis Based On PSO Optimized RBF Neural Network

As an important source of nuclear energy technology,the radioactivity is generally existed in the field mine,and the exploration of uranium resources has been facing the technology blockade of Western developed countries.In uranium exploration,gamma spectrum well is one of the important technical means,The radioactivity in uranium mine is detected,collected,and the obtained data analysis is performed by the detector,and the obtained ? energy spectrum can be obtained from the content of the radionuclide and the content of the nucleation in the mine.Traditional gamma energy spectrum analysis often requires a series of processing processes such as smooth,peak,peak boundary division,this deduction,decomposition overlapping peak,nuclide identification,coefficient correction,and each step in the process directly affects the qualitative Quantitative analysis accuracy.The traditional gamma energy spectrum is usually optimized in the process,and the accuracy of its analysis has a large error,and its analysis process is cumbersome,complicated,the amount of calculation,and the required professional qualities require high requirements.In order to overcome the defects and lack of traditional gamma energy analysis,this paper proposes the analysis of uranium ? energy spectroscopy data based on improving RBF neural network.The steps in the process of the traditional gamma spectrum quantitative analysis process are cumbersome,and the amount of calculation is high.The RBF Neural Network Model is simple to process data,and only the simple data pretreatment of the gamma energy spectrum data is required,and the uranium content can be performed.This effectively reduces the impact of complex formation environment on the statistics of gamma energy,and effectively retains an important feature of gamma energy.Improper parameters selection in the RBF neural network model may have a large problem of quantitative analysis.This paper is improved by parameter improvement to the RBF neural network.For the PSO algorithm,it is easy to fall into the defects of local optimal and premature convergence,and improve the inertial weights and learning factors in the particle group algorithm and perform performance testing.The improved PSO algorithm is combined with the RBF neural network.In the data analysis of the optimized RBF neural network to the ? energy spectrum,the radionuclide U,TH in the wild uranium ? can be quantitatively analyzed,and a certain degree of contrast is carried out.The results show that the average relative error rate of the U content value and the laboratory chemical analysis interpretation of the RBF neural network algorithm is only about 5%.By comparing the standard background model,the model quantitative analysis is faster,and has strong generalization ability.It has better practicality.