基于径向基神经网络仿真γ能谱模板库的核素识别方法

Radionuclide identification method based on a gamma-spectra template library simulated by radial basis function neural networks

传统的γ能谱分析方法存在计算复杂、耗时较长等问题。该文通过Geant4软件模拟生成26种放射性核素γ能谱,并基于径向基神经网络进行拟合,建立γ能谱模板库。针对未知γ能谱,利用最小二乘法、非线性规划算法在模板库中寻找放射性核素的最优组合,利用集成学习的思想,集成两种算法计算结果,并建立客观的识别标准。运用所提方法识别单核素γ能谱、双核素混合γ能谱以及三核素混合γ能谱,识别结果表明该方法识别核素种类的准确率较高,具有可行性与有效性。

The traditional gamma ray spectrum analysis methods are usually computationally complex and time-consuming.This paper simulated 26 radionuclide spectra using Geant4 to develop a gamma-spectra template library by fitting the spectra with radial basis function neural networks.Unknown gamma ray spectra were then identified using a least-squares algorithm and a nonlinear programing algorithm to find the optimal combination of radionuclide spectra in the library that matched the unknown spectrum with ensemble learning used to integrate the results of the two algorithms for identification.A single spectrum and mixed spectra containing 2 or 3 kinds of radionuclides were generated to evaluate the method.The results show that this method can accurately identify the radionuclides in an efficient and effective way.