机器学习法优化硅微条探测器电荷重建  

Optimization of Charge Reconstruction with Machine Learning Methods for the Silicon Micro-Strip Detectors

作　　者：闵江洪 乔锐[2] 于龙昆 彭文溪[2] 龚轲[2] 郭东亚[2] 崔兴柱[2] 刘雅清[2] MIN Jiang-hong;QIAO Rui;YU Long-kun;PENG Wen-xi;GONG Ke;GUO Dong-ya;CUI Xing-zhu;LIU Ya-qing(School of Information Engineering,Nanchang University,Nanchang 330031,China;Institute of High Energy Physics,Chinese Academy of Sciences,Beijing 100049,China;School of Advanced Manufacturing Engineering,Nanchang University,Nanchang 330031,China)

机构地区：[1]南昌大学信息工程学院,南昌330031 [2]中国科学院高能物理研究所,北京100049 [3]南昌大学先进制造学院,南昌330031

出　　处：《核电子学与探测技术》2023年第2期318-324,共7页Nuclear Electronics & Detection Technology

基　　金：国家自然科学基金天文联合基金项目(U1738133)资助。

摘　　要：为了消除硅微条探测器的电荷响应对入射位置和入射方向的依赖,提高电荷分辨能力,使用OPTICS、高斯混合与主成分分析三种机器学习算法协同完成了电荷重建,并与经典方法进行对比。结果表明,相较于经典方法,基于机器学习重建的电荷谱峰谷比更大、电荷分辨更小,重建效果更好。Three machine learning algorithms,OPTICS,Gaussian mixture and principal component analysis,are introduced to improve the charge resolution of silicon micro-strip detectors by correcting the charge response dependence on the incident position and direction.Compared with the classical method,the charge spectrum using the machine learning algorithms has a larger peak to valley ratio and better charge resolution.

关 键 词：硅微条探测器 电荷重建 机器学习 

分 类 号：TL99[核科学技术—核技术及应用]