Nuclide Identification Algorithm Based On Portable NaI Spectrometer

Nuclide identification in nuclear material analysis appraisal, the security of nuclear facilities, environmental radioactivity monitoring as well as the prevention of nuclear terrorism, and many other aspects has important applications. The basic task of radionuclide identification is to determine the type of radionuclides according to the characteristic peak in the gamma energy spectrum or characteristic value of spectrum.And through the related solution spectrum algorithm for quantitative analysis of nuclides. In this paper, spectrum collected by the portable NaI gamma spectrometer instruments as the research object, using the effective ? Spectrum data processing techniques for spectral resolution.The main results of this paper are obtained:(1) Determing the coefficient of gaussian broadening, use the MCNP5 monte carlo simulation software to simulate the gamma energy spectrum response of two point sources 137 Cs and 60 Co under the different distance. By comparison with experimental data and verify the reliability of the simulation data. Provided the basis for the research of nuclide identification algorithms,so when experimental condition is insufficient, which can't to the actual measurement on a variety of energy rays,the monte carlo method can be used for any energy spectrum response.(2) To study the nuclide identification algorithm based on Na I portable gamma spectrometer, including the energy spectrum data smooth denoising analysis, peak identification and peak boundary determination, nuclide qualitative identification based on the characteristics of the peak position, nuclide quantitative analysis by use of full peak area method and linear background deduction method. And use the 133 Ba,137Cs,60 Co sources under the background of different radiation measurement, by deducting background, quantitative analysis of the source activity, and it is concluded that sunder the background NaI detector minimal detectable activity of different energy ray.(3) Introducing neural network pattern recognition thoughts into radionuclide identification, propose full spectrum recognition method. Algorithm has good fault tolerance and applicability, the algorithm based on energy spectrum of the whole shape, gamma energy spectrum each count as neural network input values, thus can make full use of the energy spectrum data information, improve the accuracy of the results.(4) In view of the full spectrum identification method of network training sample input dimension is big, proposed based on wavelet packet decomposition extract the characteristic identification method. The gamma energy spectrum as a discrete signal, the multi spectral data are decomposed by wavelet packet to get the characteristic values, normalized energy of each frequency band, and construct a characteristic vector as the neural network training samples. It greatly reduces the input dimension of neural network, and effectively extract the gamma energy spectrum characteristics, improve the training speed, also can make accurate to nuances of gamma energy spectrum recognition, it is a kind of good gamma energy spectrum recognition method.(5) Full spectrum recognition and the characteristic identification method based on neural network method, separately Respectively test the measured sample, nuclide identification rate reached 100%. And these methods does not require spectrum with peak search process, energy and efficiency scale, It Eliminates the analysis errors due to look for peak and energy scale, it greatly improves the accuracy of the ? spectrum analysis results.