1. School of Science,National University of Defense Technology,Changsha 410000,China
2. School of Systems engineering,National University of Defense Technology,Changsha 410000,China
Show less
History+
Received
Published Online
2026-04-21
2026-11-03
PDF
(1155KB)
Abstract
The problem that the weak thrust of non-cooperative satellites, owing to its small magnitude and long duration, is highly susceptible to being submerged in observation noise, making detection extremely difficult, was prevalent in key stages such as precise orbit determination of low-Earth-orbit satellites and orbit control of deep-space probes, including midcourse corrections and near-Moon braking. To address this problem, a spatio-temporal graph convolutional network(ST-GCN) model based on temporal learning and spatial learning was proposed in this paper. By constructing the ST-GCN model, convolutional neural network(CNN), graph neural network(GCN), and feature fusion are introduced, and network parameter strategies such as adaptive learning rate scheduling and dynamic graph adjacency weights were adopted to improve detection accuracy and effectively solve the difficulty in detection caused by the extremely small magnitude of satellite micro-thrust. Experiments on three groups of low Earth orbit(LEO) satellite micro-thrust simulation datasets show that compared with current popular temporal models, the ST-GCN model proposed in this paper demonstrates the best overall performance in terms of both accuracy and efficiency. This method has the potential to be transferred to deep-space exploration scenarios and can provide a technical reference for the situational awareness of non-cooperative space targets.
以往研究主要分为以下两类。其一是基于物理模型主导的轨道机动检测方法。由美国战略司令部在Space-Track.org公开的两行根数(Two Line Elements,TLE)空间物体目录,该数据能够反映卫星的历史运动情况[13]。文献[14]提出了以 TLE 两行根数为基础的卫星机动检测方法,将轨道的半长轴作为特征参数之一,从而实现对卫星面内的机动检测。文献[15]利用SGP4模型长期项对目标特征轨道参数的预报值,与TLE真值之差建立阈值门限,检测地球低轨道(Low Earth Orbit,LEO)LEO的轨道机动事件。文献[16]在测元特性分析的基础上,对测距、测速等进行观测误差建模。文献[17]基于小波分析理论,提出了固定分组法和移动开窗法两种方案,均能快速且准确地检测出地球同步轨道(GEostationary Orbit,GEO)卫星异常。目前物理模型主导的检测方法精度和可解释的优势仍在,但当加速度降至微米级、单圈次漂移约数10 m时,其模型精度、观测采样、特征维度和检测策略均不足以从TLE噪声中提取弱信号。
Wang B C, Li S, Mu J Z, et al. Research advancements in key technologies for space-based situational awareness[J]. Space: Science & Technology, 2022, 2022: 1-31.
[2]
LiuDZong & ZhangX Y. Enhancing space-based situational awareness: real-time observation of dynamic targets with MetaCooperative-Scheduling net. IEEE Transactions on Aerospace and Electronic Systems60(6), 8198–8211 (2024).
[3]
LiuA RXuX & XiongY Q. Maneuver strategies of starlink satellite based on space X-released ephemeris. Advances in Space Research74(7), 3157–3169 (2024).
[4]
EscribanoGSanjurjo-Rivo & SiminskiJ A. A filter for tracking non-cooperative low-thrust satellites using surveillance radar data. Acta Astronaut213, 694–707 (2023).
[5]
ZhangZ XShuL & ZhangK K. Orbit determination and thrust estimation for noncooperative target using angle-only measurement. Space: Science & Technology (2023), 1–18 (2023).
[6]
SingermanP GO’RourkeS & NarayananR M. Symbolic dynamics for radar target maneuver detection with high data rates. IEEE Transactions on Aerospace and Electronic Systems60(2), 1647–1659 (2024).
[7]
ZhangPWu & BaoyinH X. Real-time hybrid method for maneuver detection and estimation of non-cooperative space targets. Astrodynamics (8), 437–453 (2024).
[8]
LiTLiK & ChenL. Maneuver detection method based on probability distribution fitting of the prediction error. Journal of Spacecraft Rockets56(4), 1114–1120 (2019).
[9]
YuS XWang & ZhuT L. Maneuver detection methods for space objects based on dynamical model. Advance in Space Research68(1), 71–84 (2021).
[10]
Vaseghi S V. Advanced digital signal processing and noise reduction[M]. 4th Edition. New Jersey: John Wiley & Sons, Ltd, 2008.
[11]
GuanH JYaoZ & LiuY. Satellite orbit real-time prediction method based on Cubature Kalman Filter(CKF)algorithm. Science Technology and Engineering13(32), 9622–9625 (2013).
[12]
TangYZhang & LiX X. High-precision orbit prediction for low Earth orbit satellites based on machine learning. Geomatics and Information Science of Wuhan University50(11), 2373–2380 (2025).
[13]
BizalionHGuillot & Petita. Systematic TLE data improvement by neural network for most cataloged resident space objects. Advances in Space Research72(7), 2649–2659 (2023).
[14]
CuiWGuo & ZhangW. Spacecraft maneuver detection method based on TLE two-line element set. Chinese of Journal of Space Science44(1), 151–158 (2024).
[15]
ZhangT TBaiX & HaoJ. LEO spacecraft orbit anomaly detection based on prediction deviation. Chinese Space Science and Technology32(5), 40–46,68 (2012).
[16]
Peng L J. Research on spacecraft orbit surveillance and anomaly detection technology[D]. Changsha: National University of Defense Technology, Changsha, 2012.
[17]
Liu Y Y, Chang Z Q, He F, et al. GEO satellite anomaly identification based on wavelet analysis[J]. Journal of Geodesy and Geodynamics, 2010, 30(5): 73-76.
[18]
Long X, Yang L P, Qiao C Y, et al. Adaptive deep learning method for GEO satellite maneuver recognition[J]. Journal of Astronaut. 2025, 46(1): 51-67.
[19]
XueBChuan & XiaoP. Mining two-line element data to detect orbital maneuver for satellite. IEEE Access7, 129537–129550 (2019).
[20]
Anantha N, Battu B T, Pruthwik M. TLE-based A2C agent for terrestrial coverage orbital path planning[EB/OL]. (2025-8-14)[2026-6-20]. https: //arxiv.org/pdf/2508.10872.
[21]
Cipollone R, LeonzioT, Calabrò G, et al. An LSTM-based maneuver detection algorithm from satellites pattern of life[C]//Proceedings of 2023 IEEE 10th International Workshop on Metrology for AeroSpace(MetroAeroSpace). Milan, Italy: IEEE, 2023.
[22]
LiB ZLiu & GuJ J. Review of convolutional neural network research. Computer Era (4), 8–12,17 (2021).
[23]
Yu F, Koltun V. Multi-scale context aggregation by dilated convolutions[C]//Proceedings of International Conference on Learning Representations (ICLR). [S. l]: ICLR 2016.
[24]
Kipf T N, Welling M. Semi-supervised classification with graph convolutional networks[C]//Proceedings of International Conference on Learning Representations(ICLR). [S. l]: ICLR 2017.
[25]
Prechelt L. Early stopping—but when?[M]. Heidelberg: Springer, 2012.
Rights & permissions
The Author(s) 2026. Published by Higher Education Press. This is an Open Access article distributed under the terms of the CC BY license (https://creativecommons.org/licenses/by/4.0/).