System Transformation Correction And Artificial Neural Network Identification Of HPGe γ Fingerprint

Gamma spectrum analysis and identification technology is widely used in the field of nuclear security,such as nuclear accident monitoring,nuclear weapon verification,nuclear proliferation prevention and sensitive nuclear material identification in nuclear terrorism.The radionuclides in nuclear materials will be accompanied by the production of γ-ray in the process of decay.Because of the characteristic of γ-ray energy,γ-ray spectrum can be used to identify and identify nuclear materials.Therefore,γ-ray spectrum is also called γ-ray fingerprint or γ-fingerprint.Due to the influence of the hardware performance and external environment factors of the y spectrometer,the measured γ fingerprint often has the phenomenon of spectral line drift,deformation and even distortion,which leads to the reduction of the confidence of nuclear material identification,and even leads to the wrong identification results.In order to solve this problem,this paper proposes a new method of γ fingerprint correction based on system transformation theory.Through the correction and recognition of simulated and measured γfingerprint,the proposed correction method is verified.The specific work includes:(1)The influence of the system instability of γ-spectrometer on the result of γ-fingerprint identification.Two groups of γ fingerprints are simulated by Monte Carlo simulation:one is the γfingerprints whose nonlinear coefficient of the γ spectrometer changes while the nuclear material does not change;the other is the γ fingerprints whose parameters of the γspectrometer do not change and the nuclear material changes slightly.The artificial neural network is used to identify the two groups of simulated γ fingerprints,and the results are compared.The necessity of correction of γ fingerprints is pointed out.(2)Based on the theory of system transformation,the method of γ fingerprint correction is proposed.The deposition energy formed by the incident gamma ray in the detector and the channel address in the corresponding multichannel analyzer are regarded as the input and output of the gamma spectrometer system respectively,so the theoretical deposition gamma fingerprint spectrum in the detector is deduced according to the function theory of random variables,that is,the modified gamma fingerprint.(3)The validity and feasibility of the modified method are verified.The HPGe γ fingerprint of different very micro nuclear materials was modified and identified.Measured HPGe γ fingerprint:first,the statistical fluctuation γ fingerprint of known standard nuclear material;second,the two groups of nuclear material γ fingerprint with little difference from the known standard nuclear material.According to the membership degree of two groups of different nuclear material and known nuclear material statistical fluctuation γ fingerprints relative to known γ fingerprints,the recognition results before and after the correction are investigated,and the validity of the proposed correction method is verified.The research shows that it is necessary to modify the γ fingerprint.Due to the influence of the instability and nonlinearity of the γ-ray spectrometer,the shift and distortion of the spectrum line of the γ-ray fingerprint results in the decrease of the recognition confidence of the γ-ray fingerprint,so that the nuclear material with little difference can not be identified effectively.The modified method transforms the γ fingerprint from the trace address domain to the energy domain,thus obtaining the theoretical deposited γ fingerprint in the detector,eliminating the influence of the spectral line drift caused by the instability of the spectrometer system and the parameters of the spectrometer system(such as nonlinearity,etc.)on the γfingerprint,effectively improving the recognition confidence of nuclear materials,and effectively identifying the very different micronucleus materials.