摘要
由于运载重量和成本的限制,空间核动力系统的质量是一个关键参数。辐射屏蔽系统是空间核动力系统自重的主要贡献之一。因此,有必要研究空间核动力系统质量最小化的屏蔽优化问题。为了实现空间堆辐射屏蔽快速设计与智能优化,基于精英策略的快速非支配遗传算法(Non Dominated Sorting Genetic Algorithm II,NSGA-Ⅱ)与反向传播(Back Propagation,BP)神经网络算法相结合的智能优化方法,对Topaz-II空间堆辐射屏蔽进行优化,将屏蔽层的总质量与剂量率作为优化目标,得到辐射屏蔽方案Pareto最优解集。同时,为了实现屏蔽方案自动化选择,将NSGA-II得到的Pareto解集作为BP神经网络算法的数据库对数据进行分析学习,建立屏蔽方案智能决策支持系统。该系统可根据工程实际需求,设置基础理论模型,获得满足工程需求的空间堆屏蔽方案。最后利用Topaz-II空间堆模型对本文方法开展了数值验证研究,证明了本文方法的正确性与可行性,可为空间堆辐射屏蔽优化提供理论与技术支撑。
[Background]The mass of a space nuclear power system is a critical parameter due to limitation of loading and high cost of transportation.The radiation shielding system is one of the main contributions of the self weight of the space nuclear power system.[Purpose]This study aims at the shielding optimization problem for the mass minimization of space nuclear power system.[Methods]Firstly,the calculation model for the Topaz-II space stack was established and the shielding material was selected.Secondly,a masked multi-objective optimization model was built on the base of Non-dominated Sorting Genetic Algorithm-II(NSGA-II)to obtain an optimization scheme.Then,an intelligent decision support system for the shielding scheme to improve the optimization efficiency was established by coupling the BP neural network algorithm.Finally,the adjustment plan was obtained according to the intelligent decision support system,and verified by using Topaz-II space reactor model.[Results]Verification results show that the intelligent decision support system can be used to set up a basic theoretical model according to the actual needs of the project, and finally provide a space stack shielding scheme that meets the engineering needs. [Conclusions] The validity and feasibility of the method in this paper are proved, providing theoretical and technical support for the optimization of radiation shielding of space reactors.
作者
贺灿
邱小平
孙征
邵静
陈珍平
赵鹏程
HE Can;QIU Xiaoping;SUN Zheng;SHAO Jing;CHEN Zhenping;ZHAO Pengcheng(School of Nuclear Science and Technology,University of South China,Hengyang 421001,China;China Institute of Atomic Energy,Beijing 102413,China)
出处
《核技术》
CAS
CSCD
北大核心
2022年第5期65-75,共11页
Nuclear Techniques
关键词
智能决策
遗传算法
BP神经网络
屏蔽优化
Intelligent decision
Genetic algorithm
BP neural network
Shield optimization
