留言板

尊敬的读者、作者、审稿人, 关于本刊的投稿、审稿、编辑和出版的任何问题, 您可以本页添加留言。我们将尽快给您答复。谢谢您的支持!

姓名
邮箱
手机号码
标题
留言内容
验证码

多阶段偏差修正模型的检验质量磁场重建

刘野 侍行剑 蔡志鸣 杨文哲 李华旺

刘野, 侍行剑, 蔡志鸣, 杨文哲, 李华旺. 多阶段偏差修正模型的检验质量磁场重建[J]. 188bet网站真的吗 . doi: 10.37188/CO.2024-0181
引用本文: 刘野, 侍行剑, 蔡志鸣, 杨文哲, 李华旺. 多阶段偏差修正模型的检验质量磁场重建[J]. 188bet网站真的吗 . doi: 10.37188/CO.2024-0181
LIU Ye, SHI Xing-jian, CAI Zhi-ming, YANG Wen-zhe, LI Hua-wang. Magnetic field reconstruction at test mass using the multi-stage bias correction model[J]. Chinese Optics. doi: 10.37188/CO.2024-0181
Citation: LIU Ye, SHI Xing-jian, CAI Zhi-ming, YANG Wen-zhe, LI Hua-wang. Magnetic field reconstruction at test mass using the multi-stage bias correction model[J]. Chinese Optics. doi: 10.37188/CO.2024-0181

多阶段偏差修正模型的检验质量磁场重建

cstr: 32171.14.CO.2024-0181
基金项目: 国家重点研发计划(No. 2020YFC2200901)
详细信息
    作者简介:

    刘 野(1999—),男,江苏徐州人,博士生,主要从事航天器工程和航天器磁洁净方面的研究。E-mail:lauye@mail.ustc.edu.cn

Magnetic field reconstruction at test mass using the multi-stage bias correction model

Funds: Supported by
More Information
    Corresponding author: xxxxxx.com
  • 摘要:

    为了精确评估空间引力波探测任务中检验质量所受到的磁场波动、磁场梯度波动噪声,本文提出了多阶段偏差修正模型MSBCM对检验质量处磁场进行精确重建。在集成学习方法的基础上,本文构建了标准全连接神经网络模块和残差全连接神经网络模块作为多阶段偏差修正模型的弱预测模型,每个弱预测模型都将对前序模型的预测偏差进行修正,最终构成强预测模型,实现对检验质量处磁场的精确重建。在对LISA Pathfinder、eLISA和太极二号空间引力波探测航天器的检验质量处磁场重建实验中,MSBCM方法相比其他方法在敏感轴方向的平均相对误差最小。模拟在轨实验中,MSBCM方法重建检验质量1敏感轴方向的磁场波动和磁场梯度波动加速度噪声的均方根误差分别为1.68×10−17 (m/s2/Hz1/2)和4.00×10−17 (m/s2/Hz1/2)。此外,MSBCM在重建检验质量2敏感轴方向的磁场波动和磁场梯度加速度噪声的均方根误差仅次于距离加权法,分别为1.72×10−16 (m/s2/Hz1/2) 和2.93×10−16 (m/s2/Hz1/2),充分验证了本文提出方法在评估在轨空间引力波探测检验质量处磁场的优势。

     

  • 图 1  空间引力波探测航天器磁源、磁强计与检验质量分布

    Figure 1.  Distribution of magnetic sources, magnetometers, and test mass in the space gravitational wave detection spacecraft

    图 2  神经网络磁场重建法的基本结构

    Figure 2.  Structure of the Neural Network

    图 3  MSBCM的基本结构

    Figure 3.  Structure of the Multi-Stage Bias Correction Model (MSBCM)

    图 4  LISA Pathfinder数据集上MSBCM训练与测试

    Figure 4.  Training and Testing of MSBCM on the LISA Pathfinder Dataset

    图 5  eLISA数据集上MSBCM训练与测试

    Figure 5.  Training and Testing of MSBCM on the eLISA Dataset

    图 6  不同方法重建太极二号TM1周围磁场的热力图

    Figure 6.  Framework of image measuring system

    图 7  不同方法重建TM1和TM2敏感轴方向的磁场波动和磁场梯度波动加速度噪声功率谱

    Figure 7.  Reconstruction of Magnetic Field Fluctuations and Gradient Acceleration Noise along the Sensitive Axes of TM1 and TM2 Using Different Methods

    表  1  LISA Pathfinder磁场重建结果

    Table  1.   Magnetic field reconstruction results for LISA Pathfinder

    方法 ${\bar \varepsilon _{\boldsymbol{B}}}$/% ${\varepsilon _{{\boldsymbol{B}},max}}$/%
    |B| Bx By Bz |B| Bx By Bz
    TM1 Taylor 89.25 518.92 564.37 694.50 2745.34 573610.11 604412.84 2461076.40
    WD 126.26 646.60 637.91 798.67 3767.01 477011.43 739996.09 2450750.40
    Multipole 85.51 523.51 524.85 647.88 3452.78 467206.45 759302.88 1998534.80
    DWME 113.38 616.28 663.83 723.73 3645.20 451613.53 1097369.80 1873063.70
    XGBoost 86.30 755.93 884.03 1378.01 5161.49 440535.35 617789.70 3250015.40
    MSBCM 26.57 201.07 231.66 212.58 1055.81 131405.62 325384.40 327050.17
    TM2 Taylor 206.23 1477.86 1385.79 2117.36 6337.63 2501042.40 1602870.80 6981814.10
    WD 270.93 1749.38 1998.34 1929.93 5561.22 2664887.90 3045968.80 3634591.40
    Multipole 203.87 1434.70 1624.28 1774.25 5552.77 1994890.20 2521400.00 4399474.20
    DWME 249.19 1627.78 1920.93 1561.21 5248.44 2215714.60 2922788.10 1403577.60
    XGBoost 19.29 210.34 255.02 203.20 1305.22 438854.54 217929.15 279339.43
    MSBCM 18.32 220.39 215.91 200.62 1418.04 612493.07 140519.40 412914.94
    下载: 导出CSV

    表  2  eLISA磁场重建结果

    Table  2.   Magnetic field reconstruction results for eLISA

    方法${\bar \varepsilon _{\boldsymbol{B}}}$/%${\varepsilon _{{\boldsymbol{B}},max}}$/%
    |B|BxByBz|B|BxByBz
    Taylor0.842.503.183.546.851590.3913222.555296.21
    WD0.842.503.183.546.851590.3913222.555296.21
    Multipole1.012.983.133.646.751766.278960.395552.00
    DWME1.012.983.133.646.751766.278960.395552.00
    XGBoost1.218.6810.658.6525.0010906.8722410.274816.18
    MSBCM0.181.531.302.423.491177.872822.872453.91
    下载: 导出CSV

    表  3  太极二号磁场重建结果

    Table  3.   Magnetic field reconstruction results for Taiji-2

    方法${\bar \varepsilon _{\boldsymbol{B}}}$/%${\varepsilon _{{\boldsymbol{B}},max}}$/%
    |B|BxByBz|B|BxByBz
    TM1Taylor15.1184.79123.55207.12887.4433066.1448781.65425602.34
    WD8.8348.0699.05158.43743.9029353.6445913.87282891.87
    Multipole11.7455.34111.75185.63706.0215681.8952423.43362661.43
    DWME15.6179.26123.28201.891025.8733688.5762955.00374511.89
    XGBoost3.4229.9224.3837.60181.2819632.5722803.73104671.03
    MSBCM1.3512.8710.8214.38145.247294.006317.0830633.32
    TM2Taylor15.33115.74142.19153.31442.41108830.68185920.57155773.28
    WD8.7173.17104.46115.51216.97104259.11107441.8992245.70
    Multipole11.8274.48123.59134.94341.0944944.78118690.82131479.33
    DWME15.69113.35133.23147.43371.9388090.07153712.76144410.41
    XGBoost2.9422.8123.3823.0155.468802.7516855.7737746.41
    MSBCM1.1013.8511.7712.4823.7719169.0629637.8013418.15
    下载: 导出CSV

    表  4  不同方法重建磁场波动、梯度波动噪声的RMSE

    Table  4.   RMSE of magnetic field fluctuations and magnetic gradient fluctuations using different methods

    方法RMSE (m/s2/Hz1/2)
    $\delta {a_{\delta {B_x}}}$$\delta {a_{\delta \nabla {B_x}}}$
    TM1Taylor3.47×10−176.42×10−17
    WD2.74×10−171.28×10−16
    Multipole9.17×10−171.10×10−16
    DWME5.98×10−175.83×10−17
    XGBoost4.63×10−157.48×10−14
    MSBCM1.68×10-174.00×10-17
    TM2Taylor3.02×10−163.47×10−16
    WD9.39×10-172.42×10-16
    Multipole4.06×10−168.97×10−16
    DWME4.47×10−161.02×10−15
    XGBoost4.49×10−157.41×10−14
    MSBCM1.72×10−162.93×10−16
    下载: 导出CSV

    表  5  消融实验

    Table  5.   Ablation study

    方法弱预测模型${\bar \varepsilon _{\boldsymbol{B}}}$/%
    Bx (TM1)Bx (TM2)
    MSBCMStMLP20.7624.01
    StMLP19.6822.41
    ReMLP16.0114.58
    ReMLP13.8512.87
    模型一StMLP20.7624.01
    StMLP19.6822.41
    StMLP-LM15.2414.92
    StMLP-LM13.4914.44
    模型二StMLP20.7624.01
    StMLP19.6822.41
    StMLP17.2117.48
    StMLP15.5514.35
    下载: 导出CSV
    Baidu
  • [1] ABBOTT B P, ABBOTT R, ABBOTT T D, et al. Observation of gravitational waves from a binary black hole merger[J]. Physical Review Letters, 2016, 116(6): 061102. doi: 10.1103/PhysRevLett.116.061102
    [2] WANNER G. Space-based gravitational wave detection and how LISA Pathfinder successfully paved the way[J]. Nature Physics, 2019, 15(3): 200-202. doi: 10.1038/s41567-019-0462-3
    [3] SALA L. Residual test mass acceleration in LISA Pathfinder: in-depth statistical analysis and physical sources[D]. Trento: University of Trento, 2023.
    [4] ZHANG H Y, XU P, YE Z Q, et al. A systematic approach for inertial sensor calibration of gravity recovery satellites and its application to Taiji-1 mission[J]. Remote Sensing, 2023, 15(15): 3817. doi: 10.3390/rs15153817
    [5] 宋长孝, 于信, 白素平, 等. 光轴稳定探测系统无热化光机结构设计[J]. 中国光学,2024,17(4):909-920. doi: 10.37188/CO.2023-0226

    SONG CH X, YU X, BAI S P, et al. Design of athermalization optical machine structure for optical axis stability detection system[J]. Chinese Optics, 2024, 17(4): 909-920. (in Chinese). doi: 10.37188/CO.2023-0226
    [6] 郭蕊香, 贾晓军, 谢常德, 等. 实用化多功能光压缩器[J]. 物理学报,2002,51(6):1262-1267. doi: 10.3321/j.issn:1000-3290.2002.06.022

    GUO R X, JIA X J, XIE CH D, et al. Compact nonclassical light source——“Squeezer”[J]. Acta Physica Sinica, 2002, 51(6): 1262-1267. (in Chinese). doi: 10.3321/j.issn:1000-3290.2002.06.022
    [7] CAÑIZARES P, CONCHILLO A, GARCÍA-BERRO E, et al. The diagnostics subsystem on board LISA pathfinder and LISA[J]. Classical and Quantum Gravity, 2009, 26(9): 094005. doi: 10.1088/0264-9381/26/9/094005
    [8] MATEOS I, DIAZ-AGUILÓ M, GESA L, et al. Magnetic field measurement using chip-scale magnetometers in eLISA[C]. Journal of Physics: Conference Series, IOP Publishing, 2015: 012028.
    [9] MATEOS I, RAMOS-CASTRO J, LOBO A. Low-frequency noise characterization of a magnetic field monitoring system using an anisotropic magnetoresistance[J]. Sensors and Actuators A: Physical, 2015, 235: 57-63. doi: 10.1016/j.sna.2015.09.021
    [10] DIAZ-AGUILÓ M, GARCÍA-BERRO E, LOBO A. Theory and modelling of the magnetic field measurement in LISA PathFinder[J]. Classical and Quantum Gravity, 2010, 27(3): 035005. doi: 10.1088/0264-9381/27/3/035005
    [11] MATEOS I, DÍAZ-AGUILÓ M, RAMOS-CASTRO J, et al. Interpolation of the magnetic field at the test masses in eLISA[J]. Classical and Quantum Gravity, 2015, 32(16): 165003. doi: 10.1088/0264-9381/32/16/165003
    [12] LIU B B, YANG ZH, QIANG L E, et al. Magnetic field recovery technique based on distance weighting multipole expansion method[J]. Europhysics Letters, 2023, 143(5): 59003. doi: 10.1209/0295-5075/acf51f
    [13] SPANTIDEAS S T, CAPSALIS C N. Validation of a source identification method for prediction of low-frequency magnetic fields in space missions[J]. IEEE Magnetics Letters, 2018, 9: 1-5.
    [14] TROUGNOU L. Ecss space systems electromagnetic compatibility handbook[C]. 2012 ESA Workshop on Aerospace EMC, IEEE, 2012: 1-6.
    [15] TSATALAS S, VERGOS D, SPANTIDEAS S T, et al. A novel multi-magnetometer facility for on-ground characterization of spacecraft equipment[J]. Measurement, 2019, 146: 948-960. doi: 10.1016/j.measurement.2019.07.016
    [16] JACKSON J D. Classical Electrodynamics[M]. New York: Wiley, 1999.
    [17] 柴国志, 黄亮, 乔亮, 等. 星上剩磁对惯性传感器的影响[J]. 中国光学,2019,12(3):515-525. doi: 10.3788/co.20191203.0515

    CHAI G ZH, HUANG L, QIAO L, et al. Effect of the on-board residual magnetism on inertial sensors[J]. Chinese Optics, 2019, 12(3): 515-525. (in Chinese). doi: 10.3788/co.20191203.0515
    [18] VITALE S. Effect of eddy currents on down-conversion of magnetic noise[J]. Memo LISA Technology Package, 2007. (查阅网上资料, 不确定本条文献类型及格式是否正确, 请确认) .
    [19] TROUGNOU L. Ac magnetic susceptibility of LisaPF test masses[R]. ESTEC, 2007. (查阅网上资料, 未找到本条文献出版地, 请确认) .
    [20] FERTIN D, TROUGNOU L. LisaPF test masses acceleration noise due to eddy currents[R]. ESTEC, 2007. (查阅网上资料, 未找到本条文献出版地, 请确认) .
    [21] YANG F C, BAI Y ZH, HONG W, et al. A charge control method for space-mission inertial sensor using differential UV LED emission[J]. Review of Scientific Instruments, 2020, 91(12): 124502. doi: 10.1063/5.0013232
    [22] SUMNER T J, MUELLER G, CONKLIN J W, et al. Charge induced acceleration noise in the LISA gravitational reference sensor[J]. Classical and Quantum Gravity, 2020, 37(4): 045010. doi: 10.1088/1361-6382/ab5f6e
    [23] TASKINEN J, YLIRUUSI J. Prediction of physicochemical properties based on neural network modelling[J]. Advanced Drug Delivery Reviews, 2003, 55(9): 1163-1183. doi: 10.1016/S0169-409X(03)00117-0
    [24] DIAZ-AGUILO M, LOBO A, GARCÍA–BERRO E. Neural network interpolation of the magnetic field for the LISA Pathfinder Diagnostics Subsystem[J]. Experimental Astronomy, 2011, 30(1): 1-21. doi: 10.1007/s10686-011-9215-8
    [25] XU L. A proportional differential control method for a time-delay system using the Taylor expansion approximation[J]. Applied Mathematics and Computation, 2014, 236: 391-399. doi: 10.1016/j.amc.2014.02.087
    [26] SHEVTSOVA I. On the accuracy of the approximation of the complex exponent by the first terms of its Taylor expansion with applications[J]. Journal of Mathematical Analysis and Applications, 2014, 418(1): 185-210. doi: 10.1016/j.jmaa.2014.03.075
    [27] LU G Y, WONG D W. An adaptive inverse-distance weighting spatial interpolation technique[J]. Computers & Geosciences, 2008, 34(9): 1044-1055.
    [28] ALAEE R, ROCKSTUHL C, FERNANDEZ-CORBATON I. An electromagnetic multipole expansion beyond the long-wavelength approximation[J]. Optics Communications, 2018, 407: 17-21. doi: 10.1016/j.optcom.2017.08.064
    [29] HE K M, ZHANG X Y, REN SH Q, et al. Deep residual learning for image recognition[C]. Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition, IEEE, 2016: 770-778.
    [30] RANGANATHAN A. The levenberg-marquardt algorithm[J]. Tutoral on LM algorithm, 2004, 11(1): 101-110. (查阅网上资料, 未找到本条文献卷期页码, 请确认) .
    [31] CHEN T Q, GUESTRIN C. XGBoost: a scalable tree boosting system[C]. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, Association for Computing Machinery, 2016: 785-794.
    [32] MATEOS I, DIAZ-AGUILÓ M, GIBERT F, et al. Magnetic back action effect of magnetic sensors for eLISA/NGO[C]. 9th International LISA Symposium, ASP, 2013: 341-345.
    [33] STONE E C, FRANDSEN A M, MEWALDT R A, et al. The advanced composition explorer[J]. Space Science Reviews, 1998, 86(1-4): 1-22.
    [34] KESKAR N S, SOCHER R. Improving generalization performance by switching from Adam to SGD[J]. arXiv: 1712.07628, 2017. (查阅网上资料, 不确定本条文献类型及格式是否正确, 请确认) .
  • 加载中
图(7) / 表(5)
计量
  • 文章访问数:  85
  • HTML全文浏览量:  31
  • PDF下载量:  5
  • 被引次数: 0
出版历程
  • 收稿日期:  2024-10-05
  • 录用日期:  2025-01-21
  • 网络出版日期:  2025-01-22

目录

    /

    返回文章
    返回
    Baidu
    map