A Flight Trajectory Prediction Method Based on Deep Learning with Attention Mechanism
Keywords:
Trajectory prediction, sequence-to-sequence, gated recurrent unit (GRU), attention mechanism, deep learningAbstract
This paper addresses the critical challenges in data-driven trajectory prediction for high-speed vehicles, focusing on issues such as training instability, computational inefficiency, and the mismatch between input and output sequence lengths. To overcome these challenges, we propose an attention-augmented Gated Recurrent Unit (GRU) sequence-to-sequence (Seq2Seq) framework that integrates temporal attention mechanisms to selectively emphasize informative historical states. This enhancement enables robust long-horizon trajectory predictions based on limited observational data. The proposed model synergizes the parameter efficiency and reduced complexity of GRUs with the dynamic focus capabilities provided by attention mechanisms, resulting in improved prediction accuracy without imposing significant computational burdens—thereby making the approach well-suited for real-time deployment on resource-constrained platforms. Comparative evaluations against baseline models using Long Short-Term Memory (LSTM) Seq2Seq and conventional GRU Seq2Seq architectures without attention demonstrate a substantial reduction in trajectory prediction errors. Extensive simulation results confirm kilometer-level prediction accuracy, validating both the effectiveness and practical viability of the presented method for high-speed vehicle trajectory prediction.
References
[1] Cheng GZ, Fu XN. Tracking algorithm of hypersonic vehicle based on two-factor enhancement. Dynam Syst Control 2020; 9(2): 81-98.
https://doi.org/10.12677/DSC.2020.92008
[2] Luo Y, Tian X, Wang H, et al. Trajectory prediction of hypersonic vehicles based on control quantity prediction. In: Proc IEEE Int Conf Information Technology, Networking, Electronic and Automation Control (ITNEC); 2020 Jun 12-14; Chongqing, China.
https://doi.org/10.1109/ITNEC48623.2020.9084956
[3] Hu JC, Zhang J, Chen WC. Analytical solutions of steady glide trajectory for hypersonic vehicle and planning application. J Beijing Univ Aeronaut Astronaut 2016; 42: 961-968.
[4] Yu WB, Chen WC. Entry guidance with real-time planning of reference based on analytical solutions. Adv Space Res 2015; 55(9): 2325-2345.
https://doi.org/10.1016/j.asr.2015.02.002
[5] Yu W, Yang J, Chen W, et al. Analytical trajectory prediction for near-first-cosmic-velocity atmospheric gliding using a perturbation method. Acta Astronaut 2021; 187: 79-88.
https://doi.org/10.1016/j.actaastro.2021.06.030
[6] Zeng L, Zhang HB, Zheng W. A three-dimensional predictor-corrector entry guidance based on reduced-order motion equations. Aerosp Sci Technol 2018; 73: 223-231.
https://doi.org/10.1016/j.ast.2017.12.009
[7] Mostafa K, Abdel MS, Ahmed MY. Flight simulation and drag prediction for a pitching-accelerating hypersonic reentry vehicle. J Spacecr Rockets 2023; 60(1): 230-238.
https://doi.org/10.2514/1.A35441
[8] Xue X, Huang S, Wei D.An adaptive multivariate Student’s t-process recursive method for hypersonic glide vehicle trajectory prediction[J].IET Radar Sonar & Navigation, 2023, 17(6): 1061-1077.
https://doi.org/10.1049/rsn2.12400
[9] Xiao CH, Li J, Lei HM, et al. Near-space target tracking based on improved IMM-UKF algorithm. J Detect Control 2018; 3: 108-113.
[10] Li JL, Guo J, Tang SJ. Trajectory prediction based on multi-model and multi-intent fusion for hypersonic gliding targets. J Astronaut 2024; 45(2): 167-180.
[11] Kuai JW, Zhao KX, Sun LG, et al. A method of space debris reentry time prediction using LSTM neural network based on ballistic coefficient pre-estimation. J Astronaut 2022; 43(12): 1731-1738.
[12] Zuo X, Zhu R, Zhou YK. Online tracking and prediction of slip ring degradation using chaos theory-based LSTM neural network. Meas Sci Technol 2023; 34(5): 055012.
https://doi.org/10.1088/1361-6501/acb5b6
[13] Khan M, Wang H, Riaz A, et al. Bidirectional LSTM-RNN-based hybrid deep learning frameworks for univariate time series classification. J Supercomput 2021; 77: 7021-7045.
https://doi.org/10.1007/s11227-020-03560-z
[14] Florent A, Arnaud LF. An LSTM network for highway trajectory prediction. In: Proc IEEE Intell Transp Syst Conf (ITSC); 2017 Oct 16-19; Yokohama, Japan.
[15] Sun LH, Yang BQ, Ma J. Trajectory prediction in pipeline form for intercepting hypersonic gliding vehicles based on LSTM. Chin J Aeronaut 2023; 36(5): 421-433.
https://doi.org/10.1016/j.cja.2023.02.017
[16] Cai YL, Deng YF, Su YH. LSTM-based trajectory classification and prediction method for hypersonic vehicles. In: Proc 21st China Conf System Simulation Technology and Its Applications; 2020. p.303-307.
[17] Li MJ, Zhou CJ, Lei HM, et al. An intelligent trajectory prediction algorithm of reentry glide target based on control parameter estimation. Syst Eng Electron 2023; 43(1): 221-233.
[18] Dong X, Tian Y, Dai L, Li J, Wan L. A new accurate aircraft trajectory prediction in terminal airspace based on spatio-temporal attention mechanism. Aerospace 2024; 11(9): 718.
https://doi.org/10.3390/aerospace11090718
[19] Xu Y, Pan Q, Wang ZF, Hu BQ. A novel trajectory prediction method based on CNN, BiLSTM, and multi-head attention mechanism. Aerospace 2024; 11(10): 822.
https://doi.org/10.3390/aerospace11100822
[20] Zhang Z, Guo D, Zhou S, et al. Flight trajectory prediction enabled by time-frequency wavelet transform. Nat Commun 2023; 14: 5258.
https://doi.org/10.1038/s41467-023-40903-9
[21] Cho K, van Merriënboer B, Gulcehre C, et al. Learning phrase representations using RNN encoder-decoder for statistical machine translation. In: Proc Conf Empir Methods Nat Lang Process (EMNLP); 2014. p.1724-1734.
https://doi.org/10.3115/v1/D14-1179
[22] Sutskever I, Vinyals O, Le QV. Sequence to sequence learning with neural networks. In: Adv Neural Inf Process Syst 2014; 27: 3104-3112.
[23] Zhang ZC, Cai YL, Fang YZ. An attention-based Seq2Seq model for predicting the boost-phase trajectory of ballistic missiles. In: Proc 40th Chinese Control Conf (CCC); 2021. p.553-558.
[24] Bao CY, Zhou X, Wang P, et al. A deep reinforcement learning-based approach to onboard trajectory generation for hypersonic vehicles. Aeronaut J 2023; 127: 1638-1658.
https://doi.org/10.1017/aer.2023.4
[25] Zhang J, Xiong J, Li L, et al. Motion state recognition and trajectory prediction of hypersonic glide vehicle based on deep learning. IEEE Access 2022; 10: 21095-21108.
https://doi.org/10.1109/ACCESS.2022.3150830
[26] Berg M de, van Kreveld M, Overmars M, Schwarzkopf O. Computational geometry: algorithms and applications. 3rd ed. Santa Clara: Springer; 2008.
