对美国国家点火装置2010年以来实验设计思路的分析

Analysis of design principles of the experiments on the National Ignition Facility since 2010

美国国家点火装置(NIF)自2010年投入使用以来,已经进行了约1030发次的惯性约束聚变研究实验.在经历了最初7年多的艰难探索之后,自2017年以来,激光聚变反应输出能量接连突破55 kJ和170 kJ,特别是在2021年8月的实验中, NIF研究团队获得了1.35MJ聚变输出能量的结果,已经接近实现靶点火(target ignition)的门槛. NIF实验数据具有极高的分析价值,近些年来NIF研究团队已经将这些数据用于进一步实验的优化设计、预测产额、矫正模拟等目的.由于NIF实验数据库中大量数据未被公开,我国科研工作者只能从少量已公开数据中了解其实验历程,无法深入分析各阶段NIF实验及各时间节点NIF团队对下一阶段实验设计思路的来源.本文根据NIF实验数据的特点,采用预测平均匹配方法和信赖域方法对NIF实验缺失数据进行了数据还原研究,并且对还原数据进行了可靠性验证.利用还原数据,本文分析了过去十年间不同阶段NIF实验的不同侧重点以及设计思路,并且利用机器学习方法预测了未来NIF实验中的热斑压强.这些结果为我国科研工作者持续跟进并深入理解最新NIF实验结果提供了一种可行的方法,也可以对我国激光聚变点火实验的设计起到借鉴作用.

Since completion of the National Ignition Facility(NIF)in 2010,more than 1030 experiments were carried out to achieve ignition.Though the experiments were unsuccessful in the first 8 years,the NIF has improved the experimental designs and achieved fusion yields from 55kJ,170kJ to 1.35MJ since 2019,approaching to the ignition milestone.The designs are based on the experimental database,which has been widely used for optimization design,yield prediction,corrected simulation,etc.However,so far the published experimental data is very limited.Also,it is difficult to obtain a completion data matrix for analyzing and understanding the experimental designs of NIF experiments at each stage and to know how the NIF sets strategic priorities for each phase.In this paper,we proposed an optimization method,which combines the PMM algorithm and trust region algorithm,to restore the missing NIF experimental data.Based on the completed data,the design principles of experiments on the NIF were analyzed,and the hot spot pressure was predicted by machine learning algorithms.The results may be helpful for the designs of laser fusion ignition experiments in China.