A Study For Decomposition Method Of Lower-energy Resolution Gamma-ray Spectra Data

This paper is accomplished based on a number of research projects: “A new method of real-time nuclear spectrum data processing based on the heterogeneous embedded systems”(This work is supported by Natural Science Foundation from Jiangxi Province(No.20114BAB211026)), “Research on the Real-Time Improvement and Application of the Embedded ARM9-Linux Scheduling Policy”(This work is supported by Jiang Xi Province Department of Education 2010 Fund for young scientist(No.GJJ10178)), “Research on Spectrum Data Processing and Spectra Decomposition Technology in Embedded Multi-channel Gamma-Ray Spectrometry”(this work is supported by Open Science Fund from Key Laboratory of Radioactive Geology and Exploration Technology Fundamental Science for National Defense,(No.2010RGET11)), “Research on the Gamma Ray Spectral Characteristics Analysis and Monte Carlo Simulation of La Br3 Detector”(this work is supported by Engineering Research Center of Nuclear Technology Application, Ministry of Education).Na I(Tl) scintillation detector has some excellent characteristic such as high detection efficiency, high light yield and low cost. So it is widely applied in many fields. However, due to the limited energy resolution of the detector, overlap among similar spectral peaks results in difficulties on searching peak. Furthermore, the Compton scattering effect of Gamma photon in Na I(Tl) crystal makes large number of low energy gamma-ray component stack in the low-energy region. So total of low-energy gamma-ray is increased and the boundary of spectral peak becomes indistinctive on the low energy region. In addition the rate of radionuclide identification is lowered under the high radiation background, or even misjudge. Thus the complexity of the corresponding spectral processing algorithm is increased significantly when samples which include more different radionuclides and complex spectrum are analyzed. In order to solve radionuclide qualitative and quantitative analysis confusion which is caused by complexity of the extracting spectral parameters because of limited energy resolution of Na I(Tl) scintillation detector, some energy-spectra high-resolution decomposition methods are proposed to solve the problem.These methods include peak search based on deconvolution, background subtraction based on adaptive for Full Width at Half Maximum, peak net area calculation based on Gauss-Newton method, response matrix method. By building on the source and the γ-ray spectrum between the Monte Carlo response matrix and Gaussian response matrix, combined with the algorithm Gold and improved Boosted-Gold algorithm, extracting various parameters of overlapping peaks, On this basis, for each nuclear event generating parameter is fast and accurate resolution under the corresponding spectral response matrix. In academic aspect, the new methods will contribute to developing theory of intelligent nuclear instrument and it will be widely applied in mineral exploration.Compared with detection technology and data analysis methods about limited energy resolution gamma spectrometry, innovative work is carried out. The decomposition method of complex instrument-spectrum is studied and verified:(1) A resolution method of γ-spectrum based on Monte-Carlo response matrix is proposed. Radionuclide can emit gamma photons and form the instrument spectrum through Na I(Tl) detector. Monte Carlo method is applied to simulate Na I(Tl) detector response to gamma photons. Combined with the Gold algorithm and improved Boosted-Gold algorithm, response matrix which is obtained by an integrated simulation and interpolation algorithm is used to implement efficient analysis of gamma spectroscopy. And the algorithm is applied to gamma energy spectrum which mixture of 238U-232 Th series, 238 U series,232 Th series, 40 K, 137 Cs and 60 Co anlysed. The results show that in the algorithm, some complicated process methods such as the peak searching, background subtraction and overlapping peaks decomposition can be omitted while qualitative and quantitative analysis can be directly used. Furthermore, the improved Boosted Gold algorithm can accelerate the decomposition of the measured spectrum and the characteristic peak distribute in a few narrow channels.(2)The unite peak search method based on deconvolution and boundary determination method based on least square fitting of gaussian function with weight factor ‘W’ is proposed.Compared with the various peak searching and boundary determination methods such as derivative, symmetric zero-area, the full-width and Gaussian function fitting, for the unite peak search method based on deconvolution and various peak boundary determinations, an evaluation system is proposed under Matlab platform. At the same time, because the traditional gaussian fitting determining the boundary of peak is not ideal, least square fitting of gaussian function with weight factor ‘W’ is proposed to determine boundary. The practice indicates that this method is stability. For peak boundary determination of single or combination of peak, good convergence can be obtained.(3) A background subtraction method of matching Full Width at Half Maximum(FWHM)---Self-adaptive SNIP(Statistics Sensitive Nonlinear Iterative Peak-Clipping Algorithm) is proposed.The gamma-ray instrument spectra,detected by Na I(Tl) detector, is composed of full-energy peak and scattering background.Therefore, in order to obtain the net count rate of radioactive sources, background count rate from the measured energy spectrum count rate and natural background radiation must be subtracted. The method can effectively subtract background to reduce influence of the scattering peak, background peak and the characteristic X-ray peak on calculating peak net areas of full-energy peak. Compared with the original SNIP algorithm, improved adaptive SNIP method can set the value of key parameter m effectively. And a good result in the automatic spectral decomposition is achieved.(4) A resolution method of γ-spectrum based on Gaussian response matrix is proposed. According to physical characteristics of forming spectra and detector features,and different energy gamma photons in the Na I(T1) detector response to the different FWHM of photoelectric peak. Then statistical properties of spectra show that the photoelectric peak can be fitted by Gaussian function. Therefore, a response matrix of the Gaussian function based on the different FWHM is proposed.Then by deconvolution, the response matrix can analyze experimental spectra which include nuclides of 238 U series, 232 Th series, 40 K accurately. Compared with the response matrix from Monte Carlo’s simulation interpolation, Gaussian response matrix has few computations on decomposing. So long as FWHM is measured precisely, more effective decompose spectrum is gain. Furthermore, under different circumstances, the method is available after background abstraction of gamma spectroscopy. The decomposing based on Gaussian response matrix has wider prospects.