基于机器学习的低轨卫星高精度轨道预报

唐宇, 张伟, 李星星, 付元辰, 张柯柯

唐宇, 张伟, 李星星, 付元辰, 张柯柯. 基于机器学习的低轨卫星高精度轨道预报[J]. 武汉大学学报 ( 信息科学版). DOI: 10.13203/j.whugis20230411
引用本文: 唐宇, 张伟, 李星星, 付元辰, 张柯柯. 基于机器学习的低轨卫星高精度轨道预报[J]. 武汉大学学报 ( 信息科学版). DOI: 10.13203/j.whugis20230411
TANG Yu, ZHANG Wei, LI Xingxing, FU Yuanchen, ZHANG Keke. High-Accuracy Orbit Prediction of Low Earth Orbit Satellites Using Machine Learning Algorithms[J]. Geomatics and Information Science of Wuhan University. DOI: 10.13203/j.whugis20230411
Citation: TANG Yu, ZHANG Wei, LI Xingxing, FU Yuanchen, ZHANG Keke. High-Accuracy Orbit Prediction of Low Earth Orbit Satellites Using Machine Learning Algorithms[J]. Geomatics and Information Science of Wuhan University. DOI: 10.13203/j.whugis20230411

基于机器学习的低轨卫星高精度轨道预报

基金项目: 

国家重点研发计划(2021YFB2501102)。

详细信息
    作者简介:

    唐宇,本科生,研究方向为低轨卫星精密定轨。ytangSGG@whu.edu.cn

    通讯作者:

    李星星,博士,教授。xxli@sgg.whu.edu.cn

High-Accuracy Orbit Prediction of Low Earth Orbit Satellites Using Machine Learning Algorithms

  • 摘要: 高精度低轨卫星预报轨道是全球低轨导航增强应用的基础。目前应用较为广泛的动力学外推方法存在着受动力学模型精度影响大、预报误差累积快的问题。为弥补这一缺陷,将机器学习方法与动力学外推法结合以提升低轨卫星轨道预报精度,并对支持向量机(support vector machine,SVM)、反向传播(back propagation,BP)神经网络和长短期记忆(long short term memory,LSTM)神经网络3种机器学习方法在低轨卫星轨道预报中的适用性进行了深入分析。利用Sentinel-3A卫星2019-01-01—2019-07-14的数据集进行了预报时长为180 min的短期轨道预报实验,对模型最优参数进行了探究,基于各模型最优参数配置进行轨道预报实验。结果表明,三种机器学习方法均能有效提升低轨卫星轨道预报精度,提升效果在不同预报时长下存在明显差异。当轨道预报时间为20 min,LSTM神经网络模型对预报轨道的精度提升效果(38.1%)显著优于SVM(-0.1%)与BP(-1.2%),预报轨道精度可达1.07 cm(3D)。当预报时长大于40 min,SVM和BP模型效果优于LSTM模型。在预报时长达到180 min时SVM和BP模型仍可以实现6-7 cm的轨道预报精度,相比于传统动力学法提升了约50%。
    Abstract: Objectives: Precise orbit prediction of low earth orbit (LEO) satellites is the foundation of LEO augmentation navigation. Due to the impact of non-conservative perturbation forces, conventional dynamic extrapolation method relies on high-accuracy dynamic models and suffer from rapid error accumulation. Methods: We combine machine learning algorithms and dynamic integration method to improve the accuracy of LEO orbit prediction. Three machine learning models are analyzed, including support vector machine (SVM), back propagation (BP) and long short term memory (LSTM) neural networks. To compare the performance of different machine learning algorithms in the high-accuracy orbit prediction, we process the orbit error time series of Sentinel-3A from January 1, 2019 to July 14, with the dynamic features selected by extreme gradient boosting (XGBoost). The optimal input configuration parameters for each machine learning models are further investigated to exploit the potential of the models in the orbit prediction. Results: The result shows that the performance of three machine learning models in the prediction varies from prediction time. When the prediction time is limited to 20 min, LSTM model presents the best performance with an accuracy improvement of 38.1%, which is evidently superior to SVM (-0.1%) and BP (-1.2%), and the corresponding orbit accuracy is 1.07 cm (3D). For the prediction exceeding 40 min, SVM and BP models outperform LSTM. When the prediction time reaches 180 min, the predicted orbit can still achieve the accuracy around 6-7 cm with the employment of SVM/BP models. Conclusion: Three models can effectively improve the orbit prediction accuracy of LEO satellites. LSTM model can achieve the best performance in ultra-short prediction (20 min), while SVM and BP models are more suited to the prediction exceeding 40 min. Compared with the traditional orbit prediction method, the improvement of machine learning models is about 40-50% overall.
  • [1]

    Reid T, Neis A, Walter T, et al. Leveraging Commercial Broadband LEO Constellations for Navigating [C]. Proceedings of the 29th International Technical Meeting of the Satellite Division of the Institute of Navigation (ION GNSS +2016). Portland, Oregon:ION, 2016:2300-2314.

    [2]

    Ge H, LI B, GE M, et al. Initial Assessment of Precise Point Positioning with LEO Enhanced Global Navigation Satellite Systems (LeGNSS) [J]. Remote Sensing, 2018, 10(7):984.

    [3]

    Li X, Ma F, Li X, et al. LEO Constellation-augmented Multi-GNSS for Rapid PPP Convergence[J]. Journal of Geodesy, 2019, 93(5):749-764.

    [4] Zeng Tian, Sui Lifen, Jia Xiaolin, et al. Simulation of Combined Orbit Determination with a Small LEO Constellation and BDS-3 Full Constellation[J]. Geomatics and Information Science of Wuhan University, 2022, 47(1):61-68. (曾添, 隋立芬, 贾小林, 等. 小型化LEO星座与BDS-3全星座联合定轨仿真[J]. 武汉大学学报(信息科学版), 2022, 47(01):61-68.)
    [5]

    Teng Y, Jia X, Peng G. LEO Navigation Augmentation Constellation Design and Precise Point Positioning Performance Analysis Based on BDS-3[J]. Advances in Space Research, 2023, 72(06):1944-1960.

    [6] Chang Zhiqiao, Hao Jinming, Zhang Chengjun. Precision Analysis of Orbit Prediction of GPS Satellites[J]. Engineering of Surveying and Mapping, 2006(02):27-29+39.(常志巧, 郝金明, 张成军. GPS精密星历的外推精度分析[J]. 测绘工程, 2006,(02):27-29+39.)
    [7]

    Ge H, Li B, Ge M, et al. Improving Low Earth Orbit (LEO) Prediction with Accelerometer Data[J]. Remote Sensing, 2020, 12(10):1599.

    [8]

    Peng H, Bai X. Artificial Neutral Network-based Machine Learning Approach to Improve Orbit Prediction Accuracy[J]. Journal of Spacecraft and Rockets, 2018, 55(05):1248-1260.

    [9]

    Peng H, Bai X. Improving Orbit Prediction Accuracy Through Supervised Machine Learning[J]. Advances in Space Research, 2018, 61(10):2628-2646.

    [10] Cao Lei. Satellite Orbit Prediction Method Based on Compensation Model by Using Deep Neural Network[D]. Nanjing:Nanjing University of Aeronautics and Astronautics, 2015. (曹磊. 基于深度神经网络补偿模型的轨道预报技术[D]. 南京:南京航空航天大学, 2015.)
    [11] Yang Xianrui. Research on Satellite Orbits Predicting Algorithm Based on Deep Learning[D]. Harbin:Harbin Institute of Technology, 2020. (杨先睿. 基于深度学习的卫星轨道预报算法研究[D]. 哈尔滨:哈尔滨工业大学, 2020.)
    [12] Zhang Xinyu, Liu Yuan, Song Jianing. Short-term Orbit Prediction Based on LSTM Neural Network[J]. Journal of Systems Engineering and Electronics, 2022, 44(03):939-947. (张心宇, 刘源, 宋佳凝. 基于LSTM神经网络的短期轨道预报[J]. 系统工程与电子技术, 2022, 44(03):939-947.)
    [13] Sun Yixuan, Shao Chunfu, Ji Xun, et al. Urban Traffic Accident Tine Series Prediction Model Based on Combination of ARIMA and Information Granulation SVR[J]. Journal of Tsinghua University (Science and Technology), 2014, 54(03):348-353+359. (孙轶轩, 邵春福, 计寻, 等. 基于ARIMA与信息粒化SVR组合模型的交通事故时序预测[J]. 清华大学学报(自然科学版), 2014, 54(03):348-353+359.)
    [14]

    Rumelhart D, McClelland J. Parallel Distributed Processing[M]. Volsland2, MIT Press, 1986.

    [15] Chen Zhigao, Wu Zihao, Ban Ya, et al. Discharge Prediction in Tidal Reach Using Harmonic Analysis and VMD-BP Neural Network[J]. Geomatics and Information Science of Wuhan University, 2023, 48(08):1389-1397. (陈志高, 吴子豪, 班亚, 等. 基于调和分析及VMD-BP神经网络的感潮河段流量预报[J]. 武汉大学学报(信息科学版), 2023, 48(08):1389-1397.)
    [16]

    Hochreiter S, Schmidhuber J. Long Short-Term Memory[J]. Neural Computation, 1997, 9(08):1735-1780.

    [17] Yang Li, Wu Yuxi, Wang Junli, et al. Research on Recurrent Neural Network[J]. Journal of Computer Applications, 2018, 38(S2):1-6+26. (杨丽, 吴雨茜, 王俊丽, 等. 循环神经网络研究综述[J]. 计算机应用, 2018, 38(S2):1-6+26.)
    [18]

    [EB/OL].(2016-01-14) [2023-10-17].https://www.esa.int/Applications/Observing_the_Earth/Copernicus/Sentinel-3/Rise_and_shine_for_Sentinel-3A

    [19]

    Li X, Han X, Li X, et al. GREAT-UPD:An Open-source Software for Uncalibrated Phase Delay Estimation Based on Multi-GNSS and Multi-frequency Observations[J]. GPS Solutions, 2021, 25:1-9.

    [20] Yang Qing, Wang Chenwei. A Study on Forecast of Global Stock Indices Based on Deep LSTM Neural Network[J]. Statistical Research, 2019, 36(03):65-77. (杨青, 王晨蔚. 基于深度学习LSTM神经网络的全球股票指数预测研究[J]. 统计研究, 2019, 36(03):65-77.)
    [21] Li Zhanshan, Liu Zhaogeng, Ding Guoxuan, et al. Feature Selection Algorithm Based on XGBoost[J]. Journal on Communications, 2019, 40(10):101-108. (李占山, 刘兆赓, 丁国轩, 等. 基于XGBoost的特征选择算法[J]. 通信学报, 2019, 40(10):101-108.)
    [22] Wang Deyun, Zhang Ludan, Wu Qi, et al. Flood Risk Assessment Based on Machine Learning Algorithms:A Case Study of Yichang City[J]. Resources and Environment in the Yangtze Basin, 2023, 32(08):1710-1723.(王德运, 张露丹, 吴祈, 等. 基于机器学习算法的洪涝灾害风险评估——以宜昌市为例[J]. 长江流域资源与环境, 2023, 32(08):1710-1723.)
    [23]

    The European Space Agency. Near Real-Time, Slow-Time Critical or Non-Time Critical?[EB/OL].[2023-10-17].https://sentinels.copernicus.eu/web/sentinel/user-guides/sentinel-3-altimetry/product-types/nrt-or-ntc

计量
  • 文章访问数:  690
  • HTML全文浏览量:  43
  • PDF下载量:  177
  • 被引次数: 0
出版历程
  • 收稿日期:  2024-02-27
  • 网络出版日期:  2024-03-13

目录

    /

    返回文章
    返回