| 57 | 0 | 49 |
| 下载次数 | 被引频次 | 阅读次数 |
为了进一步提升北斗三号全球卫星导航系统（BDS-3）超快速轨道预报精度，提出一种带有空间嵌入的采样卷积交互网络（SCINet-SE）算法：基于深度学习，提取超快速轨道历史预报误差的时空特征，并对之后一段时间的预报误差进行预测和补偿。实验选取几组不同类型的深度学习模型，除SCINet-SE外，还包括基于时间卷积网络（TCN）的采样卷积交互网络（SCINet），以及基于循环神经网络（RNN）的分段循环神经网络（SegRNN）、长短期记忆神经网络（LSTM）和双向长短期记忆网络（BiLSTM），分别测试以上几种模型对BDS-3现有的中圆地球轨道（MEO）与倾斜地球同步轨道（IGSO）超快速轨道预报的改进效果。结果表明，在对历元间隔为15 min的BDS-3超快速轨道的1 d、7 d和15 d预报改进中，SCINet-SE对测试的所有27颗卫星的轨道预报误差三维均方根（3D RMS）平均改进率可分别达到19.18%、17.11%和14.43%，均为所有测试模型中的最优值；SCINet-SE能够有效提升BDS-3(MEO、IGSO)超快速轨道预报精度。
Abstract：In order to further improve the prediction accuracy of the ultra-rapid orbits of BeiDou-3 global navigation satellite system(BDS-3), the paper proposed a sample convolution and interaction network algorithm with spatial embedding(SCINetSE): the spatiotemporal features of the historical prediction errors of ultra-rapid orbits were extracted based on the deep learning, and then, subsequent prediction errors were predicted and compensated; finally, several different types of deep learning models were selected for the experiment, including SCINet based on temporal convolutional network(TCN), and segment recurrent neural network(SegRNN), long short-term memory(LSTM) neural network and bidirectional LSTM(BiLSTM) network based on recurrent neural network(RNN), in addition to SCINet-SE, and the improvement effects of the above models on the ultra-rapid orbit prediction of BDS-3's existing medium Earth orbit(MEO) and inclined geosynchronous orbit(IGSO) were separately tested. Results showed that for 1 d, 7 d and 15 d BDS-3 ultra-rapid orbit prediction improvement with 15 min intervals, SCINet-SE could achieve optimal average three-dimensional root mean square(3D RMS) error improvement rates of 19.18%, 17.11% and 14.43%, respectively, across all 27 tested satellites, outperforming all other tested models. In conclusion, the proposed algorithm could effectively improve the prediction accuracy of BDS-3(MEO, IGSO) ultra-rapid orbits.
[1]李星星,张伟,袁勇强,等. GNSS卫星精密定轨综述:现状、挑战与机遇[J].测绘学报, 2022, 51(7):1271-1293.
[2]戴金倩,戴小蕾,彭亚权,等.联合SLR的BDS精密定轨光压模型影响与分析[J].导航定位学报, 2023,11(5):1-11.
[3]李冉,胡小工,唐成盼,等.北斗卫星导航系统混合星座的光压摄动建模和精度分析[J].武汉大学学报(信息科学版), 2018, 43(7):1063-1070.
[4]阮仁桂,贾小林,冯来平,等.北斗三号MEO卫星非保守力建模[J].测绘学报, 2022, 51(9):1862-1869.
[5]李晓杰,潘玲,郭睿,等.基于补偿波形调整的导航卫星轨道预报方法[J].武汉大学学报(信息科学版), 2017,42(8):1061-1067.
[6]李晓杰,郭睿,黄金,等.神经网络在北斗导航卫星轨道预报中的应用[J].武汉大学学报(信息科学版), 2015,40(9):1253-1258.
[7]崔东东,吕广涵,张恒璟,等.一种CEEMD-TCN的BDS轨道误差预报算法[J].导航定位学报, 2022,10(2):76-84.
[8]周建华,杨龙,徐波,等.一种导航卫星中长期轨道预报方法[J].测绘学报, 2011, 40(9):39.
[9]HOCHREITER S, SCHMIDHUBER J. Long short-term memory[J]. Neural Computation, 1997, 9(8):1735-1780.
[10]ELMAN J L. Finding structure in time[J]. Cognitive Science, 1990, 14(2):179-211.
[11]ROSENBLATT F. The perceptron:A probabilistic model for information storage and organization in the brain[J].Psychological Review, 1958, 65(6):386.
[12]GLOROT X, BENGIO Y. Understanding the difficulty of training deep feedforward neural networks[C]//Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics. JMLR Workshop and Conference Proceedings. Chia Laguna Resort, Sardinia, Italy:[s.n.], 2010:249-256.
[13]唐宇,张伟,李星星,等.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), 2025,50(11):2373-2380. DOI:10.13203/j.whugis20230411.
[14]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(3):939-947.
[15]吉长东,杨超,王强.基于EEMD-LSTM模型的BDS卫星轨道预报精度分析[J].大地测量与地球动力学,2023, 43(4):345-350.
[16]吉长东,张萌,王强.北斗卫星轨道预报方法分析[J].测绘科学, 2020, 45(7):18-25.
[17]LIU M, ZENG A, CHEN M, et al. Scinet:Time series modeling and forecasting with sample convolution and interaction[J]. Advances in Neural Information Processing Systems, 2022, 35:5816-5828.
[18]李明晓,张恒才,仇培元,等.一种基于模糊长短期神经网络的移动对象轨迹预测算法[J].测绘学报, 2018,47(12):1660-1669.
[19]ZHOU H, ZHANG S, PENG J, et al. Informer:Beyond efficient transformer for long sequence time-series forecasting[C]//PROCEEDINGS of the AAAI Conference on Artificial Intelligence. Palo Alto, California, USA:AAAI Press, 2021, 35(12):11106-11115.
[20]LECUN Y, BOSER B, DENKER J S, et al. Backpropagation applied to handwritten zip code recognition[J]. Neural Computation, 1989, 1(4):541-551.
[21]HE K, ZHANG X, REN S, et al. Deep residual learning for image recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas, NV, USA:IEEE Computer Society, 2016:770-778.
[22]BAI S, KOLTER J Z, KOLTUN V. Trellis networks for sequence modeling[C]//International Conference on Learning Representations. Vancouver, BC, Canada:OpenReview.net, 2018.
[23]AL SHALABI L, SHAABAN Z, KASASBEH B. Data mining:A preprocessing engine[J]. Journal of Computer Science, 2006, 2(9):735-739.
[24]SRIVASTAVA N, HINTON G, KRIZHEVSKY A, et al. Dropout:A simple way to prevent neural networks from overfitting[J]. The Journal of Machine Learning Research, 2014, 15(1):1929-1958.
[25]HEWAGE P, BEHERA A, TROVATI M, et al. Temporal convolutional neural(TCN)network for an effective weather forecasting using time-series data from the local weather station[J]. Soft Computing, 2020, 24:16453-16482.
[26]LIN S, LIN W, WU W, et al. SegRNN:Segment recurrent neural network for long-term time series forecasting[J].ArXiv Preprint ArXiv:2308.11200, 2023.
[27]潘雄,黄伟凯,赵万卓,等.基于BiLSTM模型的BDS-3短期钟差预报精度研究[J].测绘学报, 2024, 53(1):65-78.
[28]SCHUSTER M, PALIWAL K K. Bidirectional recurrent neural networks[J]. IEEE Transactions on Signal Processing, 1997, 45(11):2673-2681.
[29]ESTER M, KRIEGEL H P, SANDER J, et al. A density-based algorithm for discovering clusters in large spatial databases with noise[C]//Proceedings of the Second International Conference on Knowledge Discovery and Data Mining(KDD-96). Portland, Oregon, USA:AAAI Press, 1996:226-231.
[30]RUDER S. An overview of gradient descent optimization algorithms[J]. ArXiv Preprint ArXiv:1609.04747, 2016.
[31]WILLMOTT C J, MATSUURA K. Advantages of the mean absolute error(MAE)over the root mean square error(RMSE)in assessing average model performance[J]. Climate Research, 2005, 30(1):79-82.
基本信息:
DOI：10.16547/j.cnki.10-1096.20260101
中图分类号:P228.4;V412.41
引用信息:
[1]谢胜达,李建文.SCINet-SE:BDS-3超快速轨道预报改进算法[J].导航定位学报,2026,14(01):1-10.DOI:10.16547/j.cnki.10-1096.20260101.
2025-03-12
2025
2025-04-22
2025
1
2026-03-04
2026-03-04
2026-03-04