The Study Of ? Spectrum Analysis And Radioactive Nuclide Identification Method

With the widespread use of nuclear technology in field like industry,agriculture,medical and scientific research,the threat of nuclear materials diffusion,radiation terrorist attacks,and radiation accident is increasingly more serious,raised new requirement for nuclear safety and nuclear detection technology.This paper centered on the practical problem met in applications of nuclear technology,study the method of ? spectrum analysis and identification,mainly solved the ? spectrum analysis and multi-nuclide identification with strong noise and low resolution.Main content includes the following aspects:Preprocessing the energy spectrum signal.On the basis of analyzing the formation of ? spectrum,the traditional analytical method based on full peak information is used to analyze the measured and simulated ? spectra.Adopt the Least Squares method and Markov method complete the smoothing of noise spectrum while retained ? spectral important features of the peak.Through differential method and "Quantum Sphere" method to find the peak of the spectrum,can improve the ability to distinguish peaks,and realized the feature extraction of ? spectra.In view of the defect about the traditional analytical method that parsing effect is unclear for ? spectrum with high background and low resolution,proposed a full spectrum analysis method based on singular value decomposition,has a good distinction for low resolution ? spectrum and can avoid the effect of artificial set parameters to the identification result.In addition,it's hard to complete the analysis of overlap peak through traditional analytical method,against to this problem,used non-negative matrix factorization method,can be conducive to the analysis of overlap peak.Researching the application of SVM in ? spectrum identification field,aiming at the specific requirements in nuclide identification proposed some adjustment.Through analysis of the measured spectral data,discuss identification results of single nuclide and mixed nuclide.This method can complete the identification of low-resolution energy spectrum and effectively identify the composition of mixed nuclide samples.