An iterative regularization method is proposed for neutron spectrum unfolding of the liquid organic scintillator EJ-301. The Arnoldi iteration is used to generate Krylov subspaces. GMRES (Generalized Minimal Residual) combined with Tikhonov regularization is used for neutron spectrum updating. The restart method is applied to speed up the convergence as well as non-negativity constraints. The Monte Carlo code Geant4 is applied to simulate the response functions and PHSs (pulse height spectrums) of the neutron spectrums for unfolding. Four typical neutron spectrums are studied, such as the mono-energetic DD fusion neutron spectrum, ²⁴¹Am-Be neutron spectrum, ²⁵²Cf neutron spectrum and Gaussian broadened DT fusion neutron spectrum. Comparison study between the proposed method and the existing methods is also carried out, such as GRAVEL, MLEM, and direct Tikhonov regularization. Unfolding results with acceptable accuracies are obtained with the proposed method, and the proposed method shows better adaptabilities than the other three methods with both the ideal and simulated PHSs in most cases.

针对液态有机闪烁体EJ-301的中子谱展开，提出了一种迭代正则化方法。Arnoldi 迭代用于生成 Krylov 子空间。GMRES（Generalized Minimal Residual）结合Tikhonov正则化用于中子谱更新。重新启动方法用于加速收敛以及非负约束。应用蒙特卡罗代码 Geant4 来模拟展开中子谱的响应函数和 PHS（脉冲高度谱）。研究了四种典型的中子谱，如单能DD聚变中子谱、²⁴¹ Am-Be中子谱、²⁵²Cf 中子谱和高斯展宽 DT 聚变中子谱。还进行了所提出方法与现有方法的比较研究，例如GRAVEL、MLEM和直接Tikhonov正则化。所提出的方法获得了具有可接受精度的展开结果，并且在大多数情况下，所提出的方法比其他三种方法在理想和模拟的PHSs下表现出更好的适应性。