原子能科学技术2025，Vol.59Issue(3)：700-707,8.DOI:10.7538/yzk.2024.youxian.0524
基于遗传算法的快中子靶站束流引出优化设计
Optimization Design of Fast Neutron Target Station Beam Extraction Based on Genetic Algorithm
摘要
Abstract
An accelerator neutron source target station for energy calibration using the time-of-flight(TOF)method was designed in this paper.Considering the special requirements of high-repetition-rate proton pulses from the accelerator and the off-target facility(OTF)method for calibration,it is necessary to reduce the impact of scattering on the trajectory of neutrons generated by the target to obtain the highest possible proportion of uncollided neutrons.At the same time,to achieve a lightweight and compact structure for the target station,constraints were also placed on the dose rate around the measurement point,mainly referring to the neutron flux values outside the diameter range of 5 m toϕ50 mm from the compound target during the optimization process.During calibration,the intensity of neutrons will also affect the overall efficiency,so the value of the uncollided neutron flux is also an important design indicator.The distance between the target and the collimator,the material thickness and structure of the collimator,and the inner diameter of the vacuum pipeline are variable parameters that affect the aforementioned indicators.Therefore,the aforementioned parameters are used as variables,and the three important indicators are taken as the objective function.The structural composition of the neutron extraction pipeline was optimized through theoretical simulation,coupled with a genetic algorithm and Monte Carlo code.In the optimization design,the genetic algorithm was used to initialize and adjust the optimization parameters,generate Monte Carlo input files,and call the Monte Carlo code for calculation.The total number of iterations is 20 000(200×100),with each calculation taking approximately 1 minute.The Monte Carlo code can simulate physical processes such as neutron,photon,proton,and coupled neutron-photon transport,and record physical quantities such as uncollided neutron flux,total neutron flux,and neutron flux around the measurement point.The software simulation uses the ENDF/B-Ⅶdatabase for cross-sections.By repeatedly calling the Monte Carlo code and employing excellent evolutionary strategies,the genetic algorithm can ensure that the optimization design evolves in a converging direction.The results show that the optimized scheme reduces the total thickness of the collimator material by about 30%compared to the initial structure,the inner diameter of the vacuum pipeline is reduced by 7.17 cm,and the proportion of uncollided neutrons increases by 0.8%.After overall structural optimization,the surrounding dose rate is below 1 μSv/h,meeting the requirements for dose protection.The neutron flux at 5 m after shielding and collimation is uniformly distributed within a diameter range ofϕ50 mm.关键词
靶站设计/遗传算法/蒙特卡罗模拟/屏蔽准直器/飞行时间法Key words
target station design/genetic algorithm/Monte Carlo simulation/shielding collimator/time-of-flight method分类
能源科技引用本文复制引用
覃子倩,谭新建,张小东,潘清泉,刘霄..基于遗传算法的快中子靶站束流引出优化设计[J].原子能科学技术,2025,59(3):700-707,8.基金项目
国家自然科学基金(12305190,12075192) （12305190,12075192）
