Study Of Age Identification Of Simulated Nuclear Materials

In order to safeguard world peace and stability,nuclear weapons in the word must be strictly controlled.It's very important to master advanced nuclear techniques for a nation to protect its safety interest and has a fair chance to participate in nuclear disarmament.Nuclear material age is one of the most important attributes of concern in the verification of nuclear weapons,therefore,it's quite significant to explore age identification techniques of nuclear materials.?-rays can be given off during the radioactive decay in nuclear materials,and ? spectra is unique for a nuclear material,so the ? spectra is defined as “radiation fingerprints” which can be used to identify the nuclear materials containing different radio-nuclides of different radio-activities.In this work,radioactive sources ⁶⁰ Co ? ¹⁵² Eu and ¹³⁷ Cs are chosen as the simulated nuclear materials,and firstly,? spectra of their combinations were simulated by Monte-Carlo method as the radiation fingerprints with different ages.Secondly,the wavelet low frequency coefficients of the ? spectra were extracted with wavelet transform as the features for identifying the ages of the simulated nuclear materials.Finally,the ages of the simulated nuclear materials were identified quantitatively with RBF artificial neural network through establishing the training sample set.The main aim of this work is to establish the optimal training sample set which has minimum size and can ensure the accuracy.The main research contents are as follows:?1?Influence of half-life of nuclide on the sampling period of training samples;?2?Influence of the number of ?-rays from nuclear materials on the sampling period of training samples;?3?Influence of sampling range on the sampling period of training samples;?4?Influence of the ratio between sampling range and half-life on the sampling period of training samples;?5?Age identification of simulated nuclear materials containing different nuclides.The results showed that for the nuclear material of single radionuclide,half-life of the radionuclide?sampling range and the ratio between the two has significant influences on the sampling period.In a certain sampling range,the smaller the half-life of the radionuclide,the smaller the maximum sampling period.When the sampling range is much larger than the half-life of radionuclide,the sampling period should be small.The sampling period can besignificantly increased when the sampling range is close to or less than the half-life of radionuclide.When the sampling period is constant,the error increases with the increase of the sampling range.The number of the ?-rays from the nuclear materials has no influence on the selection of training sample,so it can be ignored.For the nuclear materials containing a variety of different radio-nuclides,the whole sampling range of interest must be divided into two sections according to minimum half-life,and the training sample sets must be established separately for the two different sampling ranges with different sampling period.In a word,it's effective and feasible to identify quantitatively the nuclear material age by RBF network through selecting the reasonable sampling ranges and sampling period,and establishing optimal training sample sets.