EI、Scopus 收录
中文核心期刊

留言板

尊敬的读者、作者、审稿人, 关于本刊的投稿、审稿、编辑和出版的任何问题, 您可以本页添加留言。我们将尽快给您答复。谢谢您的支持!

姓名
邮箱
手机号码
标题
留言内容
验证码

基于LSTM模型的飞行器智能制导技术研究

汪韧 惠俊鹏 俞启东 李天任 杨奔

汪韧, 惠俊鹏, 俞启东, 李天任, 杨奔. 基于LSTM模型的飞行器智能制导技术研究. 力学学报, 2021, 53(7): 2047-2057 doi: 10.6052/0459-1879-20-388
引用本文: 汪韧, 惠俊鹏, 俞启东, 李天任, 杨奔. 基于LSTM模型的飞行器智能制导技术研究. 力学学报, 2021, 53(7): 2047-2057 doi: 10.6052/0459-1879-20-388
Wang Ren, Hui Junpeng, Yu Qidong, Li Tianren, Yang Ben. Research of LSTM model-based intelligent guidance of flight aircraft. Chinese Journal of Theoretical and Applied Mechanics, 2021, 53(7): 2047-2057 doi: 10.6052/0459-1879-20-388
Citation: Wang Ren, Hui Junpeng, Yu Qidong, Li Tianren, Yang Ben. Research of LSTM model-based intelligent guidance of flight aircraft. Chinese Journal of Theoretical and Applied Mechanics, 2021, 53(7): 2047-2057 doi: 10.6052/0459-1879-20-388

基于LSTM模型的飞行器智能制导技术研究

doi: 10.6052/0459-1879-20-388
详细信息
    作者简介:

    汪韧, 工程师, 主要研究方向: 飞行器动力学与制导. E-mail: wangren94@126.com

  • 中图分类号: V412.4+4

RESEARCH OF LSTM MODEL-BASED INTELLIGENT GUIDANCE OF FLIGHT AIRCRAFT

  • 摘要: 人工智能技术的突破性进展为飞行器再入制导技术的研究提供了新的技术途径, 本文针对预测校正制导中两方面的问题: 一是纵向“预测环节”积分计算量大和“校正环节”割线法迭代求解难以满足实时性的问题, 二是纵向制导和横向制导都需要对动力学方程进行积分, 存在明显的冗余计算问题, 提出基于长短期记忆网络(long short-term memory, LSTM) 的飞行器智能制导技术. 一方面, 在纵向制导中不需要对动力学方程进行积分来预测待飞射程, 即去除“预测环节”; 另一方面, 不再基于割线法迭代求解倾侧角的幅值, 即去除倾侧角的“校正环节”, 大大减少积分计算量, 提高计算速度. 利用深度学习在神经网络映射能力和实时性方面的双重天然优势, 基于再入飞行器的实时状态信息, 采用LSTM模型实时生成倾侧角指令. 同时, 将纵向和横向制导环节的制导周期统一为一个周期, 进一步确保制导系统满足在线制导的实时性要求. 蒙特·卡罗仿真分析表明, 本文所提的方法在飞行器再入初始状态和气动参数拉偏情况下具有精度和速度上的优势.

     

  • 图  1  预测校正制导与基于LSTM模型的再入制导对比

    Figure  1.  Comparison of predictor-corrector guidance and LSTM model-based reentry guidance

    图  2  基于LSTM的神经网络模型架构

    Figure  2.  Structure diagram of LSTM-based neural network

    图  3  LSTM模型

    Figure  3.  LSTM model

    图  4  损失函数随迭代次数的变化曲线

    Figure  4.  Loss function-epoch curve

    图  5  均方根误差随迭代次数的变化曲线

    Figure  5.  RMSE-epoch curve

    图  6  高度−速度曲线对比

    Figure  6.  Comparison of height-velocity curve

    图  7  横向轨迹曲线对比

    Figure  7.  Comparison of horizontal trajectory curve

    图  8  倾侧角−速度曲线对比

    Figure  8.  Comparison of bank angle-velocity curve

    图  9  航迹角−时间曲线对比

    Figure  9.  Comparison of flight path angle-time curve

    图  10  航向角−时间曲线对比

    Figure  10.  Comparison of heading angle-time curve

    图  11  初始状态和气动参数扰动下落点经纬度的散布图

    Figure  11.  Scatter diagram of longitude and latitude under initial state error and aerodynamic parameter perturbation

    图  12  计算实时性对比分析

    Figure  12.  Comparison of computing time analysis

    表  1  飞行器再入初始点的参数范围

    Table  1.   Range of initial state parameters

    Initial state parametersRange
    ${h_0}/{\rm{km}}$$\left[ {60,75} \right]$
    ${\theta _0}{/(^ \circ) }$$\left[ { - {2^ \circ },{2^ \circ } } \right]$
    ${\phi _0}{/(^ \circ) }$$\left[ { - {2^ \circ },{2^ \circ } } \right]$
    ${V_0}/({{\rm{m}} \cdot{\rm{ s}}^{-1} })$$\left[ {5500,6200} \right]$
    ${\gamma _0}{/(^ \circ) }$$\left[ { - {3^ \circ },{3^ \circ } } \right]$
    ${\psi _0}{/(^ \circ) }$$\left[ { - {3^ \circ },{3^ \circ } } \right]$
    下载: 导出CSV

    表  2  飞行器再入初始状态和气动参数偏差

    Table  2.   Initial state error and aerodynamic parameter perturbation

    PerturbationDistributionError bound
    $\Delta r/{\rm{km}}$uniform distribution$ \pm 3$
    $\Delta \theta {/(^ \circ) }$uniform distribution$\pm {0.1 }$
    $\Delta \phi {/(^ \circ) }$uniform distribution$\pm {0.1 }$
    $\Delta V/({{\rm{m}}\cdot {\rm{s}}^{-1} })$uniform distribution$ \pm 100$
    $\Delta \gamma {/(^ \circ) }$uniform distribution$\pm {0.3 }$
    $\Delta \psi {/(^ \circ) }$uniform distribution$\pm {1 }$
    $\Delta {C_{\rm{L}}}/\% $uniform distribution$ \pm 30$
    $\Delta {C_{\rm{D}}}/\% $uniform distribution$ \pm 30$
    下载: 导出CSV
  • [1] 包为民. 航天飞行器控制技术研究现状与发展趋势. 自动化学报, 2013, 39(6): 697-702 (Bao Weimin. Present situation and development tendency of aerospace control techniques. Acta Automatica Sinica, 2013, 39(6): 697-702 (in Chinese)
    [2] Lu P. Entry guidance: A unified method. Journal of Guidance, Control, and Dynamics, 2014, 37(3): 713-728 doi: 10.2514/1.62605
    [3] Xue SB, Lu P. Constrained predictor-corrector entry guidance. Journal of Guidance, Control, and Dynamics, 2010, 33(4): 1273-1281 doi: 10.2514/1.49557
    [4] Goodfellow I, Bengio Y, Courville A. Deep learning. Cambridge, Massachusetts: MIT Press, 2016
    [5] Sutton RS, Barto AG. Reinforcement Learning: An Introduction. Cambridge, Massachusetts: MIT Press, 2011
    [6] Krizhevsky A, Sutskever I, Hinton GE. Imagenet classification with deep convolutional neural networks//Advances in Neural Information Processing Systems, 2012: 1097-1105
    [7] Simonyan K, Zisserman A. Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv: 1409.1556, 2014
    [8] He KM, Zhang XY, Ren SQ, et al. Deep residual learning for image recognition//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016: 770-778
    [9] Liu W, Anguelov D, Erhan D, et al. SSD: Single shot multi-box detector//The 14th European Conference on Computer Vision. 2016: 21-37
    [10] Bochkovskiy A, Wang CY, Liao HY. YOLOv4: optimal speed and accuracy of object detection. arXiv preprint arXiv: 2004.10934, 2020
    [11] Mnih V, Kavukcuoglu K, Silver D, et al. Human-level control through deep reinforcement learning. Nature, 2015, 518(7540): 529-533 doi: 10.1038/nature14236
    [12] Lillicrap TP, Hunt JJ, Pritzel A, et al. Continuous control with deep reinforcement learning. arXiv preprint arXiv: 1509.02971, 2015
    [13] Schulman J, Wolski F, Dhariwal P, et al. Proximal policy optimization algorithms. arXiv preprint arXiv: 1707.06347, 2017
    [14] Haarnoja T, Zhou A, Abbeel P, et al. Soft actor-critic: Off-policy maximum entropy deep reinforcement learning with a stochastic actor. arXiv preprint arXiv: 1801.01290, 2018
    [15] 程林, 蒋方华, 李俊峰. 深度学习在飞行器动力学与控制中的应用研究综述. 力学与实践, 2020, 42(3): 267-276 (Cheng Lin, Jiang Fanghua, Li Junfeng. A review on the applications of deep learning in aircraft dynamics and control. Mechanics in Engineering, 2020, 42(3): 267-276 (in Chinese)
    [16] 余跃, 王宏伦. 基于深度学习的高超声速飞行器再入预测校正容错制导. 兵工学报, 2020, 41(4): 656-669 (Yu Yue, Wang Honglun. Deep learning-based reentry predictor-corrector fault-tolerant guidance for hypersonic vehicles. Acta Armamentarii, 2020, 41(4): 656-669 (in Chinese)
    [17] Shi Y, Wang ZB. A deep learning-based approach to real-time trajectory optimization for hypersonic vehicles//AIAA Scitech 2020 Forum, 2020
    [18] Cheng L, Jiang FH, Wang ZB, et al. Multi-constrained real-time entry guidance using deep neural networks. IEEE Transactions on Aerospace and Electronic Systems, 2021, 57(1): 325-340 doi: 10.1109/TAES.2020.3015321
    [19] 李天任, 杨奔, 汪韧等. 基于 Q-Learning 算法的再入飞行器制导方法. 战术导弹技术, 2019(5): 44-49 (Li Tianren, Yang Ben, Wang Ren, et al. Reentry vehicle guidance method based on Q-learning algorithm. Tactical Missile Technology, 2019(5): 44-49 (in Chinese)
    [20] 张秦浩, 敖百强, 张秦雪. Q-learning 强化学习制导律. 系统工程与电子技术, 2020, 42(2): 414-419 (Zhang Qinhao, Ao Baiqiang, Zhang Qinxue. Reinforcement learning guidance law of Q-learning. Journal of Systems Engineering and Electronics, 2020, 42(2): 414-419 (in Chinese) doi: 10.3969/j.issn.1001-506X.2020.02.21
    [21] Gaudet B, Furfaro R, Linares R. Reinforcement learning for angle-only intercept guidance of maneuvering targets. Aerospace Science and Technology, 2020, 99
    [22] Hovell K, Ulrich S. Deep reinforcement learning for spacecraft proximity operations guidance. Journal of spacecraft and rockets, 2021, 58(2): 254-264 doi: 10.2514/1.A34838
    [23] Hovell K, Ulrich S. On deep reinforcement learning for spacecraft guidance//AIAA Scitech 2020 Forum, 2020
    [24] Gao JS, Shi XM , Cheng ZT, et al. Reentry trajectory optimization based on deep reinforcement learning//The 31th Chinese Control and Decision Conference, 2019: 2588-2592
    [25] Koch W, Mancuso R, West R, et al. Reinforcement Learning for UAV Attitude Control. ACM Transactions on Cyber-Physical Systems, 2019, 3(2): 1-21
    [26] Chai RQ, Tsourdos A, Savvaris A, et al. Six-DOF spacecraft optimal trajectory planning and real-time attitude control: A deep neural network-based approach. IEEE Transactions on Neural Networks and Learning Systems, 2020, 31(11): 5005-5013 doi: 10.1109/TNNLS.2019.2955400
    [27] 方科, 张庆振, 倪昆等. 高超声速飞行器时间协同再入制导. 航空学报, 2018, 39(5): 197-212 (Fang Ke, Zhang Qingzhen, Ni Ke, et al. Time-coordination reentry guidance law for hypersonic vehicle. Acta Aeronautics et Astronautica Sinica, 2018, 39(5): 197-212 (in Chinese)
    [28] 周宏宇, 王小刚, 单永志等. 基于改进粒子群算法的飞行器协同轨迹规划. 自动化学报: 1-8[2020-04-25]. https://doi.org/10.16383/j.aas.c190865
    [29] Hochreiter S, Schmidhuber J. Long short-term memory. Neural Computation, 1997, 9(8): 1735-1780 doi: 10.1162/neco.1997.9.8.1735
    [30] Sutskever I, Martens J, Dahl G, et al. On the importance of initialization and momentum in deep learning//International Conference on Machine Learning, 2013: 1139-1147
    [31] Glorot X, Bengio Y. Understanding the difficulty of training deep feedforward neural networks//The 13th International Conference on Artificial Intelligence and Statistics, 2010: 249-256
  • 加载中
图(12) / 表(2)
计量
  • 文章访问数:  99
  • HTML全文浏览量:  33
  • PDF下载量:  29
  • 被引次数: 0
出版历程
  • 收稿日期:  2020-11-15
  • 录用日期:  2021-05-26
  • 网络出版日期:  2021-05-26
  • 刊出日期:  2021-07-18

目录

    /

    返回文章
    返回