【摘要】 通过仿真由整块NaI(Tl)晶体、光导和PMT阵列组成的闪烁探测器对511keVγ射线的探测过程,研究用神经网络由闪烁光分布计算γ射线入射点位置的方法,并与传统的Anger公式定位算法进行比较。结果表明,神经网络方法可克服或减小探测器作用深度效应和边缘效应的影响,在探测器中心区域和边缘区域,对于γ射线垂直入射和大角度入射都能获得很好的位置分辨率。另外,还研究了局部神经网络在角度方向上的泛化能力。更多还原

【Abstract】 By simulating the detection procedure of 511 keV γ photons in the scintillating detector which consists of a bulk NaI(Tl) scintillation crystal,a glass light guide and a PMT array,a novel neural network estimator is applied to calculate the incidence point of γ photons corresponding to the output scintillating light from PMT array.Compared with traditional Anger algorithm,neural network can conquer or reduce the interaction-depth effect and edge effect,and give high spatial resolution.Meanwhile from the point of view of practical application,relationship between the generalization ability of neural network and the incidence angle is also studied in the paper.更多还原