基于多球中子谱仪的解谱算法研究现状

Status and Progress of Unfolding Algorithms for NeutronSpectrum Based on Multishpere Neutron Spectrometer

基于有限的探测器响应计数求解中子能谱是一个典型的非适定问题,高效准确求解能谱具有较大困难。针对该问题,广泛调研了基于多球中子谱仪求解中子能谱的原理与数学模型,指出中子解谱面临的最大困难在于先验信息的缺失;从利用先验信息的角度讨论了目前常见中子解谱算法的原理与特点;在如何利用先验信息的统一框架下,分析了最小二乘法、奇异值截断法、正则化方法、智能算法和压缩感知算法在中子解谱中的应用,指出不同解谱算法的本质区别在于利用先验信息种类和形式的不同。为准确求解能谱,须在解谱算法中合理地增加关于能谱的先验信息,在信号处理领域出现的新理论与新方法可被借鉴用于中子解谱问题,通过进一步挖掘利用先验信息来改善求解能谱的结果。

The multisphere neutron spectrometer has been widely used in neutron spectrum measurement because of its sensitive response for wide range of energy from thermal neutron energy to GeV.However,it is difficult to unfold spectrum using limited output of detectors since ill-posedness of the problem.Essentially,it is the lack of a priori information that prevents one from unfolding spectrum accurately,which is an underdetermined system of equations in mathematical formation.Therefore,to obtain reliable results,it is necessary to incorporate the priori information within unfolding process.In this paper,we discuss different kinds of unfolding algorithms,from an aspect of the variety of the the priori information and the ways of utilizing the priori information,including least square estimation,singular-value-truncation method,regularization method,intelligent algorithms,and compressive-sensing algorithm.We conclude that the unfolded results closely depend on the variety of the priori information.Furthermore,we suggest the novel algorithms in signal processing discipline,such as compressive sensing,are potential powerful methods for spectrum unfolding.