Study On Algorithms Of Overlapping Peaks Decomposition In Nuclear Spectrum Measurement

As the nuclear science and technology provide convenient services and clean energy,people gradually begin to pay close attention to the environment and physical damage caused by radiation.Usually,radioactive material which exists in the radiation environment can release gamma rays.Through the measurement of gamma rays,people can understand the nuclide types of radioactive material and judge nuclide content and activity,etc.However,in real measurement,there is a common phenomenon which is the influence of environment or other interference rays will lead to spectral signal overlapping.In a common gamma ray detector,NaI(Tl)detectors are widely used due to its high detection efficiency,convenient maintenance and moderate price.But NaI(Tl)detectors have limited low energy resolution of overlapped peak,which makes decomposition of overlapped peak become a problem in spectral analysis.Based on this background,according to the statistics dis tribution of gamma energy spectrum,Based on this background,according to the statistical distribution of the gamma ray spectrum,the expectation maximization method,genetic algorithm and particle swarm optimization(PSO)algorithm are used to decompose the simulated overlapping peaks by using MATLAB.In this paper,the main study and results are following several aspects:1.The paper discussed the energy spectrum and its mathematical model.Then the paper completed the simulation of the original overlapped peak spectral lines on MATLAB,according to the statistical fluctuation property of the spectrum.The simulation results could be as the research object of the follow-up algorithm and the basis of error analysis.2.For defects of expectation maximization algorithm in solving the decomposition problem of overlapping peaks takes long computation time,this paper proposed a fast algorithm,and completed the task of decomposition of overlapped peak by using the algorithm effectively.3.The paper completed the decomposition of overlapped peaks after expounded the superiority of genetic algorithm.The solution space and the solution in the space contained in overlapping peak decompos ition are regarded as chromosomes and genes in the genetic algorithm.Combined with the GA toolbox,a series of selection and genetic operation finds out the parameter combination which is the most consistent with the original overlapping peaks in the global model.4.To find the relationship between particle swarm algorithm and overlapping peak decomposition.Complete the discussion of the initial parameters,the selection of the fitness function,the particle evaluation and position transformation,the individual extreme value and the global extr eme value update work,and finally achieve a good decomposition effect.Then using this algorithm completed the 232 Th and 226 Ra nuclide decomposition of the actual overlap of the task,and the cause of error is analyzed.5.The expectation-maximization algorithm,genetic algorithms and particle swarm optimization are used to realize the decomposition of the bimodal overlap and the three peaks overlap peaks.In the double peak resolution of overlapping peaks: the minimum distance between the initial values can be decomposed,which are 17 KeV,13 KeV and 5 KeV respectively.When the initial peak is known,the maximum expectation method can be used to decompose 8KeV initial values,and the error of weight and standard deviation of the genetic algorithm are reduced.In the three peak resolution of overlapping peaks: the correlation peak and the deviation of the maximum expected value can improve the accuracy of expectation-maximization algorithm and genetic algorithm;the particle swarm algorithm,even when the initial parameters are unknown,can also completed 185 KeV,195 KeV and 203 KeV overlapping peaks decomposition,and the decomposition results are good.The three algorithms studied in this paper can achieve the decomposition of multimodal overlapping peaks which with similar energy,and have better decomposition effect.The paper also has some reference value for the practical resolution of overlapping peaks decomposition.