Real-time automatic detection of signals triggered by space dust's impact based on deep learning
-
摘要:
准确快速地检测航天器上发生的尘埃碰撞事件能帮助我们更好地了解空间环境的尘埃分布以及减少航天器因尘埃碰撞受到的破坏.现有人工识别或基于尘埃碰撞引起的电势差信号波形特征的机器识别尘埃碰撞事件的方法虽然有较高精度，但效率低下，迫切需要高精度且自动化方法识别航天器收集的海量电势差信号.深度学习模型在信号分类和识别具有较强能力，本文把空间尘埃碰撞引起的电势差信号检测问题建模成信号分类问题，构建了一个卷积神经网络模型，该模型可以自动提取信号特征并根据特征对信号分类，同时为了训练模型和测试模型预测准确率，构建了一个由尘埃碰撞引起的电势差信号和非尘埃碰撞引起的电势差信号组成的数据集，模型在训练集上准确率为99.46%，在测试集上准确率达到98.68%，查全率为99.44%，查准率为97.95%，threat score为97.41%.实现了高精度且自动化的尘埃碰撞事件检测.
Abstract:Accurate and rapid detection of dust impact events on spacecraft can help us better understand the dust distribution in the space and reduce the damage to spacecraft due to dust impacts. Although the existing methods of manual identification or machine identification of dust impact events based on the waveform characteristics of potential difference signals caused by dust impacts have high accuracy, their efficiency is low, and high-precision and automated methods are urgently needed to identify the massive potential difference signals collected by spacecraft. The deep learning model has strong ability in signal classification and recognition. In this paper, the problem of potential difference signals caused by dust impacts detection is modeled as a signal classification problem, and a convolutional neural network model is constructed, which can automatically extract signal features and classify signals according to the features. At the same time, in order to train the model and test the prediction accuracy of the model, a data set composed of potential difference signals caused by dust impacts and potential difference signals caused by other events was constructed. The accuracy rate of the model on training set is 99.46% and on the test set is 98.68%, the recall rate is 99.44%, the precision rate is 97.95%, and the threat score is 97.41% High-precision and automatic dust collision events detection is realized.
-
-
表 1
测试集在训练好的深度神经网络模型的预测结果
Table 1.
Predicted result on test set based on trained deep neural network
非尘埃碰撞引起的电势差信号(预测) 尘埃碰撞引起的电势差信号(预测) 非尘埃碰撞引起的电势差信号(实际) 1410 (真反例，即True Negative或TN) 30 (假正例，即False Positive或FP) 尘埃碰撞引起的电势差信号(实际) 8 (假反例，即False Negative或FN) 1432 (真正例，即True Positive或TP) 
下载: 导出CSV
-
Anders C, Bringa E M, Ziegenhain G, et al. 2012. Why nanoprojectiles work differently than macroimpactors: The role of plastic flow. Physical Review Letters, 108(2): 027601, doi:10.1103/physrevlett.108.027601.
Andersson L, Ergun R E, Delory G T, et al. 2015. The Langmuir Probe and Waves (LPW) instrument for maven. Space ScienceReviews, 195(1-4): 173-198, doi:10.1007/s11214-015-0194-3.
Asano R S, Takeuchi T T, Hirashita H, et al. 2013. Dust formation history of galaxies: a critical role of metallicity for the dust mass growth by accreting materials in the interstellar medium. Earth, Planets and Space, 65(3): 213-222, doi:10.5047/eps.2012.04.014.
Aubier M G, Meyer-Vernet N, Pedersen B M. 1983. Shot noise fromgrain and particle impacts in Saturn's ring plane. GeophysicalResearch Letters, 10(1): 5-8, doi:10.1029/GL010i001p00005.
Berera A. 2017. Space dust collisions as a planetary escape mechanism. Astrobiology, 17(12): 1274-1282, doi:10.1089/ast.2017.1662.
Braslau D. 1970. Partitioning of energy in hypervelocity impact against loose sand targets. Journal of Geophysical Research, 75(20): 3987-3999.
Calura F, Pozzi F, Cresci G, et al. 2017. The dust-to-stellar mass ratio as a valuable tool to probe the evolution of local and distant star-forming galaxies. Monthly Notices of the Royal Astronomical Society, 465(1): 54-67, doi:10.1093/mnras/stw2749.
Deng T Y, Liu E X, Xu C. 2022. SuperDARN polar ionospheric convection potential model based on deep learning. Chinese Journal of Geophysics (in Chinese), 65(3): 819-829, doi:10.6038/cjg2022P0207.
Ellerbroek L E, Gundlach B, Landeck A, et al. 2017. The footprint of cometary dust analogues-I. Laboratory experiments of low-velocity impacts and comparison with Rosetta data. Monthly Notices of the Royal Astronomical Society, 469(S2): S204-S216, doi:10.1093/mnras/stx1257.
Fischer G, Gurnett D A, Yair Y. 2011. Extraterrestrial lightning and its past and future investigation. //Wood M D ed. Lightning: Properties, Formation and Types. Hauppauge, NY: Nova Science Publishers, 19-38.
Gurnett D A, Kurth W S, Granroth L J, et al. 1991. Micron-sized particles detected near Neptune by the Voyager 2 plasma wave instrument. Journal of Geophysical Research: Space Physics, 96(S01): 19177-19186, doi:10.1029/91JA01270.
Gurnett D A, Morgan D D, Granroth L J, et al. 2010. Non-detection of impulsive radio signals from lightning in Martian dust storms using the radar receiver on the Mars express spacecraft. Geophysical Research Letters, 37(17): L17802, doi:10.1029/2010gl044368.
Guzewich S D, Talaat E R, Toigo A D, et al. 2013. High-altitude dust layers on mars: Observations with the thermal emission spectrometer. Journal of Geophysical Research: Planets, 118(6): 1177-1194, doi:10.1002/jgre.20076.
Johari D. 2017. Features of the Electric Fields Generated by Lightning with Special Attention to Positive Ground Flashes. Uppsala University Publications.
Kellogg P J, Goetz K, Monson S J. 2016. Dust impact signals on the wind spacecraft. Journal of Geophysical Research: Space Physics, 121(2): 966-991, doi:10.1002/2015JA0211.
Kellogg P J, Goetz K, Monson S J. 2018. Are STEREO single hitsdust impacts?. Journal of Geophysical Research: Space Physics, 123(9): 7211-7219, doi:10.1029/2018JA025554.
Kim J, Lee J K, Lee K M. 2016. Accurate image super-resolutionusing very deep convolutional networks. //2016 IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas, NV, USA: IEEE, 1646-1654.
Krauss C E, Horányi M, Robertson S. 2003. Experimental evidence forelectrostatic discharging of dust near the surface of mars. NewJournal of Physics, 5: 70, doi:10.1088/1367-2630/5/1/370.
Liu W, Xiao H W, Bai C C. 2019. Spatial multi-object recognition based on deep learning. //2019 IEEE International Conference on Unmanned Systems. Beijing, China: IEEE, 736-741, doi: 10.1109/ICUS48101.2019.8995980.
Melnik O, Parrot M. 1998. Electrostatic discharge in Martian dust storms. Journal of Geophysical Research: Space Physics, 103(A12): 29107-29117, doi:10.1029/98ja01954.
Meng X H, Guo L H, Zhang Z F, et al. 2008. Reconstruction of seismic data with least squares inversion based on nonuniformfast Fourier transform. Chinese Journal of Geophysics (in Chinese), 51(1): 235-241.
Meyer-Vernet N, Aubier M G, Pedersen B M. 1986. Voyager 2 at Uranus: Grain impacts in the ring plane. Geophysical Research Letters, 13(7): 617-620, doi:10.1029/GL013i007p00617.
Odelstad E. 2013. Noise Sources in the Electric Field Antenna on the ESA JUICE Satellite[Master's thesis]. Swedish Institute of Space Physics, Uppsala Division.
O'Shea E, Sternovsky Z, Malaspina D M. 2017. Interpreting dust impact signals detected by the STEREO spacecraft. Journal of Geophysical Research: Space Physics, 122(12): 11864-11873, doi:10.1002/2017ja024786.
Owens M J, Arge C N, Spence H E, et al. 2005. An event-based approach to validating solar wind speed predictions: High-speed enhancements in the Wang-Sheeley-Arge model. Journal ofGeophysical Research: Space Physics, 110(A12): A12105-n/a, doi:10.1029/2005JA011343.
Pabari J P. 2017. Detection of dust around mars and its implications. Current Science, 113(11): 2080, doi:10.18520/cs/v113/i11/2080-2084.
Pantellini F, Landi S, Zaslavsky A, et al. 2012. On the unconstrained expansion of a spherical plasma cloud turning collisionless: Case of a cloud generated by a nanometre dust grain impact on an uncharged target in space. Plasma Physics and Controlled Fusion, 54(4): 045005, doi:10.1088/0741-3335/54/4/045005.
Qiu X P. 2020. Neural Networks and Deep Learning (in Chinese). Beijing: China Machine Press.
Shao R B, Xiao L Z, Liao G Z, et al. 2022. A reservoir parametersprediction method for geophysical logs based on transfer learning. Chinese Journal of Geophysics (in Chinese), 65(2): 796-808, doi:10.6038/cjg2022P0057.
Shen M M, Sternovsky Z, Garzelli A, et al. 2021. Electrostatic model for antenna signal generation from dust impacts. Journalof Geophysical Research: Space Physics, 126(9): e2021JA029645 doi:10.1002/essoar.10507271.1.
Tarantino P, Goel A, Corso A, et al. 2018. An electrostatic method to model the expansion of hypervelocity impact plasma on positively biased surfaces. Physics of Plasmas, 25(9): 092103, doi:10.1063/1.5039656.
Wang Z, Zheng X, Li D Y, et al. 2021. A VGGNet-like approachfor classifying and segmenting coal dust particles with overlappingregions. Computers in Industry, 132: 103506, doi:10.1016/j.compind.2021.103506.
Wozniakiewicz P. 2017. Cosmic dust in space and on earth. Astronomy & Geophysics, 58(1): 1.35-1.40, doi:10.1093/astrogeo/atx027.
Yang Q J, Ren J, Xiang H. 2022. Spatio-temporal detection of auroral substorm based on deep learning. Chinese Journal of Geophysics (in Chinese), 65(3): 898-907, doi:10.6038/cjg2022P0047.
Yang Y, Shen F, Yang Z C, et al. 2018. Prediction of solar windspeed at 1AU using an artificial neural network. Space Weather, 16(9): 1227-1244, doi:10.1029/2018SW001955.
Ye S Y, Gurnett D A, Kurth W S, et al. 2014. Properties of dust particles near Saturn inferred from voltage pulses induced bydust impacts on Cassini spacecraft. Journal of Geophysical Research: Space Physics, 119(8): 6294-6312, doi:10.1002/2014ja020024.
Ye S Y, Vaverka J, Nouzak L, et al. 2019. Understanding Cassini RPWS antenna signals triggered by dust impacts. GeophysicalResearch Letters, 46(20): 10941-10950, doi:10.1029/2019gl084150.
Zhang H, Wang D N, Li H X, et al. 2017. High accurate seismic data reconstruction based on non-uniform curvelet transform. Chinese Journal of Geophysics (in Chinese), 60(11): 4480-4490, doi:10.6038/cjg20171132.
Zhao M, Chen S, Yuen D. 2019. Waveform classification and seismic recognition by convolution neural network. Chinese Journal ofGeophysics (in Chinese), 62(1): 374-382, doi:10.6038/cjg2019M0151.
邓天云, 刘二小, 徐晨. 2022. 基于深度学习的SuperDARN雷达极区电离层对流电势模型构建及预测. 地球物理学报, 65(3): 819-829, doi:10.6038/cjg2022P0207. http://www.geophy.cn/article/doi/10.6038/cjg2022P0207
孟小红, 郭良辉, 张致付等. 2008. 基于非均匀快速傅里叶变换的最小二乘反演地震数据重建. 地球物理学报, 51(1): 235-241. http://www.geophy.cn/article/id/cjg_208
邱锡鹏. 2020. 神经网络与深度学习. 北京: 机械工业出版社.
邵蓉波, 肖立志, 廖广志等. 2022. 基于迁移学习的地球物理测井储层参数预测方法研究. 地球物理学报, 65(2): 796-808, doi:10.6038/cjg2022P0057. http://www.geophy.cn/article/doi/10.6038/cjg2022P0057
杨秋菊, 任杰, 向晗. 2022. 基于深度学习的极光亚暴时-空自动检测. 地球物理学报, 65(3): 898-907, doi:10.6038/cjg2022P0047. http://www.geophy.cn/article/doi/10.6038/cjg2022P0047
张华, 王冬年, 李红星等. 2017. 基于非均匀曲波变换的高精度地震数据重建. 地球物理学报, 60(11): 4480-4490, doi:10.6038/cjg20171132. http://www.geophy.cn/article/doi/10.6038/cjg20171132
赵明, 陈石, Yuen D. 2019. 基于深度学习卷积神经网络的地震波形自动分类与识别. 地球物理学报, 62(1): 374-382, doi:10.6038/cjg2019M0151. http://www.geophy.cn/article/doi/10.6038/cjg2019M0151
-
图(4)
表(1)
计量
- 文章访问数: 3185
- PDF下载数: 154
- 施引文献: 0