空间引力波探测科学数据处理的挑战与人工智能技术的应用  

Challenges in space-based gravitational wave data analysis and applications of artificial intelligence

作　　者：王赫 杜明辉 徐鹏 周宇峰[6] WANG He;DU MingHui;XU Peng;ZHOU Yu-Feng(International Centre for Theoretical Physics Asia-Pacific,University of Chinese Academy of Sciences,Beijing 100190,China;Taiji Laboratory for Gravitational Wave Universe(Beijing/Hangzhou),University of Chinese Academy of Sciences(UCAS),Beijing 100049,China;Center for Gravitational Wave Experiment,National Microgravity Laboratory,Institute of Mechanics,Chinese Academy of Sciences,Beijing 100190,China;Hangzhou Institute for Advanced Study,University of Chinese Academy of Sciences(UCAS),Hangzhou 310024,China;Lanzhou Center of Theoretical Physics,Lanzhou University,Lanzhou 730000,China;CAS Key Laboratory of Theoretical Physics,Institute of Theoretical Physics,Chinese Academy of Sciences,Beijing 100190,China)

机构地区：[1]中国科学院大学国际理论物理中心(亚太地区),北京100190 [2]中国科学院大学引力波宇宙太极实验室(北京/杭州),北京100049 [3]中国科学院力学研究所国家微重力实验室引力波实验中心,北京100190 [4]国科大杭州高等研究院,杭州310024 [5]兰州大学兰州理论物理中心,兰州730000 [6]中国科学院理论物理前沿重点实验室,中国科学院理论物理研究所,北京100190

出　　处：《中国科学：物理学、力学、天文学》2024年第7期64-86,共23页Scientia Sinica Physica,Mechanica & Astronomica

基　　金：国家重点研发计划(编号:2021YFC2203004,2021YFC2201903);国家自然科学基金(编号:12247187,12147103);国家天文科学数据中心(编号:NADC2023YDS-01);中央高校基本科研业务费专项资金资助项目。

摘　　要：随着空间引力波探测项目,如LISA、太极、天琴等的不断推进,我们即将获得一个观察宇宙的全新视角.然而,这些项目的科学数据处理面临着前所未有的挑战,包括大量波源混叠、非稳态噪声、数据异常等.本文旨在为研究人员提供一个相对全面的视角,以贝叶斯统计推断框架为线索,综述了这些挑战及其可能的解决方案,讨论了从波源模板构建、探测器响应模拟,到噪声和数据异常处理的过程,并着重探讨了全局拟合与参数反演的技术策略,尤其是似然函数的构造与计算方法,以及如何利用多种随机采样技术提升分析效率和准确度.特别地,文章重点介绍了人工智能技术在引力波信号建模、噪声与数据异常处理、信号识别与参数估计等方面的应用,展示了人工智能如何为解决空间引力波探测数据分析中的复杂问题提供新的路径和工具.As space-based gravitational wave detection projects such as LISA,Taiji,and Tianqin continue to advance,we are on the cusp of gaining a new viewpoint on observing the universe.However,the scientific data processing for these projects faces unprecedented challenges,including the superposition of numerous gravitational wave sources,non-stationary noises,and data anomalies.This review aims to make a brief summary of these challenges and their possible solutions,using the Bayesian statistical inference framework as a thread,and provide researchers with a relatively comprehensive perspective.Topics such as the construction of waveform templates,the modeling of detector responses,and the processing of noise and data anomaly are discussed,with a focus on the strategies for parameter estimation and global fitting,especially the evaluation of likelihood,and the utilization of various stochastic sampling techniques to improve the efficiency and accuracy of analysis.Notably,this review highlights the applications of artificial intelligence technologies in waveform modeling,noise and data anomaly processing,signal recognition,and parameter estimation,showcasing how artificial intelligence can pave new paths for solving complex problems in the data analysis of space-based gravitational wave detection.

关 键 词：引力波探测 数据处理 参数估计 人工智能 

分 类 号：P111[天文地球—天文学] P142.84