真空电子技术Issue(3)：14-19,6.DOI:10.16540/j.cnki.cn11-2485/tn.2025.03.03
基于级联神经网络的电子光学系统设计
Design of Electron Optical System Based on Cascade Neural Network
摘要
Abstract
The design of electron optical systems significantly impacts the performance of vacuum devices.Due to many parameters involved,designing structures that meet specific requirements requires extensive experi-ence and considerable time.This paper proposes a method for rapid electron optical system design using a casca-ded neural network,which achieves forward prediction from electron gun structure parameters to electron beam trajectories,as well as backward design to derive electron gun structure parameters from key target electron beam trajectory parameters.During the forward prediction phase,the corresponding electron beam trajectory curve val-ues can be quickly obtained by inputting the electronic gun structure parameters.The structure parameters obtained through backward design are validated through simulations,demonstrating good compliance with the target elec-tron beam trajectory requirements.Validation analysis of the beam waist characteristics shows an average axial waist position error of approximately 6%and an average waist size error of about 2.28%.With the proposed cas-caded neural network,the design process takes only 0.25 seconds.The proposed approach enables electron gun design without the need for prior experience and facilitates rapid design.关键词
电子光学系统/级联神经网络/电子注轨迹/反向设计Key words
Electron optical system/Cascade neural network/Electron beam trajectory/Inverse design分类
信息技术与安全科学引用本文复制引用
庄永志,高宏利,洪玮,沈长圣,张长青,潘攀,冯进军,柏宁丰..基于级联神经网络的电子光学系统设计[J].真空电子技术,2025,(3):14-19,6.基金项目
国家自然科学基金(61871110),微波电真空器件国家级重点实验室开放基金(6142807012302) （61871110）
