Relativistic Navigation Method for Deep Space Probes
-
摘要:在相对论导航系统中,通过毫角秒恒星角距测量装置获取反映恒星光行差和光线引力偏折变化的恒星角距观测量,利用导航滤波器处理观测量,估计深空探测器在惯性空间的位置和速度矢量,以及敏感器的测量基准偏差。建立了面向导航滤波器设计和系统性能分析的状态方程和观测方程,根据导航系统的克拉美劳下界(Cramer-Rao Lower Bound,CRLB)考察了相对论导航方法用于深空探测器的可行性,设计了通过导航滤波器自学习提升相对论导航系统性能的策略。仿真研究表明,对于环绕火星运行的深空探测器,在恒星角距测量精度为1 mas的情况下,相对论导航方法能达到百米量级的定位精度水平。为相对论导航方法在深空中的应用提供了支持。
-
关键词:
- 深空探测/
- 天文导航/
- 相对论效应/
- Q学习扩展卡尔曼滤波器
Abstract:An autonomous navigation method based on the observations of the relativistic perturbations for deep space probes is presented in this paper. The relativistic perturbations including the stellar aberration and the starlight gravitational deflection are new type of celestial navigation measurement, which can provide the kinematic state information of the deep space probes in the inertial frame. In the relativistic navigation system, the position and velocity vectors of the deep space probes, and the measurement bias of the optical sensor can be estimated through measuring the inter-star angle perturbed by the stellar aberration and the gravitational deflection of light with an optical sensor for LOS (line-of-sight) direction with extremely high accuracy. In this paper, the state equation and measurement equation for the design of the navigation filter and the navigation performance evaluation are established. The feasibility of the relativistic navigation method for deep space probes is investigated via the calculation of the Cramer-Rao Lower Bound (CRLB). In addition, the self-learning strategy of the navigation filter is designed to enhance the relativistic navigation performance. It is illustrated through the numerical simulation that, for a Mars-circling probe, the position error of the relativistic navigation method is on the order of 100 m with the inter-star angle measurement accuracy of 1 mas.Highlights● An autonomous celestial navigation method based on the observations of the relativistic perturbations of starlight for deep space probes is presented. ● The feasibility of the presented method is investigated via the calculation of the Cramer-Rao Lower Bound (CRLB) of the relativistic navigation system. ● A novel Q-learning extended Kalman filter (QLEKF) is designed to optimize the navigation filtering parameters adaptively. -
表 1Q学习扩展卡尔曼滤波器
Table 1Q-learning extended Kalman filter
算法 1:Q学习扩展卡尔曼滤波器 输入:状态估计初始值 $ {\hat{\boldsymbol{x}}}_{0} $ 及其误差方差阵 $ {\boldsymbol{P}}_{0} $,事先确定的方差阵集合 $ \left\{{\hat{\boldsymbol{Q}}}_{k}^{(s,a)}\right\} $,观测量$ {\boldsymbol{y}}_{k} $ 1:$ {\hat{\boldsymbol{x}}}_{0}^{(s,a)}\leftarrow {\hat{\boldsymbol{x}}}_{0} $,$ {\boldsymbol{P}}_{0}^{(s,a)}\leftarrow {\boldsymbol{P}}_{0} $,$ Q(\mathrm{s},\mathrm{a})\leftarrow 0 $,$ k\leftarrow 0 $ 2:for特定导航任务周期,do 3: for任意 $ a\in \mathbf{A} $,do 4: $ R(s,a)\leftarrow 0 $ 5: for$t=\mathrm{1,2},\cdots ,T$,do 6: $ k\leftarrow k+1 $ 7: $ [{\hat{\boldsymbol{x}}}_{k}^{\left(s,a\right)},{\boldsymbol{P}}_{k}^{\left(s,a\right)},{\tilde{\boldsymbol{y}}}_{k}^{\left(s,a\right)}]\leftarrow $ $\mathrm{E}\mathrm{K}\mathrm{F}({\hat{\boldsymbol{x} } }_{k-1}^{\left(s,a\right)},{\boldsymbol{P} }_{k-1}^{\left(s,a\right)},{\boldsymbol{y} }_{k},{\hat{\boldsymbol{Q} } }_{k}^{\left(s,a\right)},{\mathit{R} }_{k})$ $\rhd $ 探索滤波器 8: $ R\left(s,a\right)\leftarrow R\left(s,a\right)+ $ $\dfrac{1}{{T} }\left\{ {\left[{\left({\tilde{\boldsymbol{y} } }_{k}^{\left(s,a\right)}\right)}^{-1}{\mathit{R} }_{k}^{-1}{\tilde{\boldsymbol{y} } }_{k}^{\left(s,a\right)}\right]}^{-1}-R\left(s,a\right)\right\}$ 9: $ [{\hat{\boldsymbol{x}}}_{k},{\boldsymbol{P}}_{k},{\tilde{\boldsymbol{y}}}_{k}]\leftarrow \mathrm{E}\mathrm{K}\mathrm{F}({\hat{\boldsymbol{x}}}_{k-1},{\boldsymbol{P}}_{k-1},{\boldsymbol{y}}_{k},{\hat{\boldsymbol{Q}}}_{k},{\mathit{R}}_{k}) $ $\rhd$ 导航滤波器 10: end for 11: $ Q\left(s,a\right)=\left(1-\alpha \right)Q\left(s,a\right)+ $$\mathrm{\alpha }[R\left(s,a\right)+\mathrm{\gamma }\underset{a' }{\mathrm{max} }Q\left(s',a'\right)]$ 12:end for 13: ${a}_{\max}\leftarrow \mathrm{arg}\underset{a}{\mathrm{max} }Q\left(s,a\right)$ 14: ${\hat{\boldsymbol{Q} } }_{k}\leftarrow {\hat{\boldsymbol{Q} } }_{k}^{\left(s,{a}_{\max}\right)}$ 15:end for 16:输出:$ \left\{{\hat{\boldsymbol{x}}}_{k}\right\} $ and $ \left\{{\boldsymbol{P}}_{k}\right\} $ 表 2扩展卡尔曼滤波器的计算公式
Table 2Equations of extended Kalman filter
算法 2:扩展卡尔曼滤波器 1:function$ \mathrm{E}\mathrm{K}\mathrm{F}({\hat{\boldsymbol{x}}}_{k-1},{\boldsymbol{P}}_{k-1},{\boldsymbol{y}}_{k},{\boldsymbol{Q}}_{k},{\boldsymbol{R}}_{k}) $ 2: ${\hat{\boldsymbol{x} } }_{k|k-1}\leftarrow{f}\left({\hat{\boldsymbol{x} } }_{k-1}\right)$ $ \rhd $ 预测 3: $ {\boldsymbol{P}}_{k|k-1}\leftarrow {\boldsymbol{F}}_{k}{\boldsymbol{P}}_{k-1}{\boldsymbol{F}}_{k}^{\mathrm{T}}+{\boldsymbol{Q}}_{k} $ 4: $ {\boldsymbol{K}}_{k}\leftarrow {\boldsymbol{P}}_{k|k-1}{\boldsymbol{H}}_{k}^{\mathrm{T}}{({\boldsymbol{H}}_{k}{\boldsymbol{P}}_{k|k-1}{\boldsymbol{H}}_{k}^{\mathrm{T}}+{\boldsymbol{R}}_{k})}^{-1} $ 5: ${\tilde{\boldsymbol{y} } }_{k}\leftarrow {\boldsymbol{y} }_{k}-{h}\left({\hat{\boldsymbol{x} } }_{k|k-1}\right)$ 6: $ {\hat{\boldsymbol{x}}}_{k}\leftarrow {\hat{\boldsymbol{x}}}_{k|k-1}+{\boldsymbol{K}}_{k}{\tilde{\boldsymbol{y}}}_{k} $ $ \rhd $ 更新 7: ${\boldsymbol{P} }_{k}\leftarrow \left({\boldsymbol I}-{\boldsymbol{K} }_{k}{\boldsymbol{H} }_{k}\right){\boldsymbol{P} }_{k|k-1}{({\boldsymbol I}-{\boldsymbol{K} }_{k}{\boldsymbol{H} }_{k})}^{\mathrm{T} }+{\boldsymbol{K} }_{k}{\boldsymbol{R} }_{k}{\boldsymbol{K} }_{k}^{\mathrm{T} }$ 8: return$ {\hat{\boldsymbol{x}}}_{k} $,$ {\boldsymbol{P}}_{k} $,$ {\tilde{\boldsymbol{y}}}_{k} $ 9:end function 表 3火星探测器初始轨道参数
Table 3Initial orbital elements of Mars probe
半长轴/km 偏心率 轨道倾角/(°) 升交点赤经/(°) 近地点幅角/(°) 平近点角/(°) 3 697 0.011 90.1549 1.647×10–5 0.3490 0.5235 表 4导航滤波器参数
Table 4Parameters of navigation filter
滤波器参数 仿真中参数设置 初始系统噪声方差阵 $ {\hat{\boldsymbol{Q}}}_{0}=\left[\begin{array}{ccc}{\sigma }_{r}^{2}{\boldsymbol{I}}_{3\times 3}& & \\ & {\sigma }_{v}^{2}{\boldsymbol{I}}_{3\times 3}& \\ & & {\sigma }_{\kappa }^{2}{\boldsymbol{I}}_{3\times 3}\end{array}\right] $,其中,$ {\sigma }_{r}=5\times {10}^{-5}\;\mathrm{m} $,$ {\sigma }_{v}=5\times {10}^{-5}\;\mathrm{m}/\mathrm{s} $,$ {\sigma }_{\kappa }=0.1\;\mathrm{m}\mathrm{a}\mathrm{s} $ 测量噪声方差阵 ${\boldsymbol{R} }_{k}={\sigma }_{a}^{2}{\boldsymbol{I} }_{3\times 3}$,其中${\sigma }_{a}=1\;\mathrm{m}\mathrm{a}\mathrm{s}$ 初始估计误差方差阵 $ {\boldsymbol{P}}_{0}=\left[\begin{array}{ccc}{p}_{r}^{2}{\boldsymbol{I}}_{3\times 3}& & \\ & {p}_{v}^{2}{\boldsymbol{I}}_{3\times 3}& \\ & & {p}_{\kappa }^{2}{\boldsymbol{I}}_{3\times 3}\end{array}\right] $,其中,${p}_{r}=1\;\mathrm{k}\mathrm{m}$,${p}_{v}=0.1\;\mathrm{m}/\mathrm{s}$,${p}_{\kappa }=0.2''$ 学习速率 $ \mathrm{\alpha }=0.1 $ 折扣因子 $ \gamma =0.9 $ 计算奖赏的周期/s $ T=500 $ 集合中的元素 对$ {\hat{\boldsymbol{Q}}}_{0} $进行放大或缩小得到$ {\hat{\boldsymbol{Q}}}_{k}^{(s,a)} $ 表 5相对论导航系统的定位测速精度
Table 5Position and velocity accuracy of relativistic navigation system
导航滤波器 EKF AEKF REKF QLEKF 平均位置误差RMS/m 348.4 198.1 553.3 163.8 平均速度误差RMS/(m·s–1) 0.29 0.17 0.47 0.11 -
[1] 房建成, 宁晓琳. 深空探测器自主天文导航方法[M]. 西安: 西北工业大学出版社, 2010. [2] 王大轶, 黄翔宇, 魏春岭. 基于光学成像测量的深空探测自主控制原理与技术[M]. 北京: 中国宇航出版社, 2012. [3] 崔平远, 高艾, 朱圣英. 深空探测器自主导航与制导[M]. 北京: 中国宇航出版社, 2016. [4] 张伟,许俊,黄庆龙,等. 深空天文自主导航技术发展综述[J]. 飞控与探测,2020,3(4):8-16.ZHANG W,XU J,HUANG Q L,et al. Survey of autonomous celestial navigation technology for deep space[J]. Fight Control & Detection,2020,3(4):8-16. [5] 徐瑞,李朝玉,朱圣英,等. 深空探测器自主规划技术研究进展[J]. 深空探测学报(中英文),2021,8(2):111-123.XU R,LI Z Y,ZHU S Y,et al. Research progress of autonomous planning technology for deep space probes[J]. Journal of Deep Space Exploration,2021,8(2):111-123. [6] CHRISTIAN J A. Optical navigation using planet's centroid and apparent diameter in image[J]. Journal of Guidance,Control,and Dynamics,2015,38(2):192-204. [7] FRANZESE V,TOPPUTO F. Optimal beacons selection for deep-space optical navigation[J]. The Journal of the Astronautical Sciences,2020,67:1775-1792.doi:10.1007/s40295-020-00242-z [8] ZHU S Y,XIU Y,ZHANG N,et al. Crater-based attitude and position estimation for planetary exploration with weighted measurement uncertainty[J]. Acta Astronautica,2020,176:216-232.doi:10.1016/j.actaastro.2020.06.033 [9] OGAWA N,TERUI F,MIMASU Y,et al. Image-based autonomous navigation of Hayabusa2 using artificial landmarks:The design and brief in-flight results of the first landing on asteroid Ryugu[J]. Astrodynamics,2020,4(2):89-103.doi:10.1007/s42064-020-0070-0 [10] ELY T A,SEUBERT J,BRADLEY N,et al. Radiometric autonomous navigation fused with optical for deep space exploration[J]. The Journal of the Astronautical Sciences,2021,68:300-325.doi:10.1007/s40295-020-00244-x [11] SHEIKH S I,PINES D J,RAY P S,et al. Spacecraft navigation using X-ray pulsars[J]. Journal of Guidance,Control,and Dynamics,2006,29(1):49-63. [12] WANG Y,ZHENG W,ZHANG D,et al. X-ray pulsar-based navigation method using one sensor and modified time-differenced measurement[J]. Proceedings of the Institution of Mechanical Engineers,Part G:Journal of Aerospace Engineering,2019,233(1):299-309.doi:10.1177/0954410017731440 [13] LIU J,NING X L,MA X,et al. Geometry error analysis in solar Doppler difference navigation for the capture phase[J]. IEEE Transactions on Aerospace and Electronic Systems,2019,55(5):2556-2567.doi:10.1109/TAES.2019.2892659 [14] NING X,CHAO W,HUANG Y,et al. Spacecraft autonomous navigation using the Doppler velocity differences of different points on the solar disk[J]. IEEE Transactions on Aerospace and Electronic Systems,2020,56(6):4615-4625.doi:10.1109/TAES.2020.3000843 [15] CHRISTIAN J A. StarNAV:autonomous optical navigation of a spacecraft by the relativistic perturbation of starlight[J]. Sensors,2019,19:4064.doi:10.3390/s19194064 [16] YUCALAN D, PECK M. A static estimation method for autonomous navigation of relativistic spacecraft[C]//Proceedings of IEEE Aerospace Conference. Big Sky, MT, USA: IEEE, 2019. [17] CALABRO E. Relativistic aberrational interstellar navigation[J]. Acta Astronautica,2011,69:360-364.doi:10.1016/j.actaastro.2011.05.013 [18] XIONG K,WEI C. Integrated celestial navigation for spacecraft using interferometer and Earth sensor[J]. Proceedings of the Institution of Mechanical Engineers,Part G:Journal of Aerospace Engineering,2020,234(16):2248-2262.doi:10.1177/0954410020927522 [19] XIONG K,WEI C,ZHOU P. Integrated autonomous optical navigation using Q-learning extended Kalman filter[J]. Aircraft Engineering and Aerospace Technology,2022,94(6):848-861.doi:10.1108/AEAT-05-2021-0139 [20] SUTTON R S, BARTO A G. Reinforcement learning: an Introduction[M]. London, England: MIT Press, 2018. [21] WEI Q,LEWIS F L,SUN Q,et al. Discrete-time deterministic Q-learning:a novel convergence analysis[J]. IEEE Transactions on Cybernetics,2017,47(5):1224-1237.doi:10.1109/TCYB.2016.2542923 [22] 熊凯, 魏春岭, 郭建新. 航天器导航滤波器设计方法[M]. 北京: 中国宇航出版社, 2022 [23] LEI M,WYK B J,QI Y. Online estimation of the approximate posterior Cramer-Rao lower bound for discrete-time nonlinear filtering[J]. IEEE Transactions on Aerospace and Electronic Systems,2011,47(1):37-57.doi:10.1109/TAES.2011.5705658 [24] RISTIC B,FARINA A,BENVENUTI D,et al. Performance bounds and comparison of nonlinear filters for tracking a ballistic object on re-entry[J]. IEE P-Radar Sonar Navigation,2003,150:65-70.doi:10.1049/ip-rsn:20030212 [25] DONG G,ZHU Z H. Autonomous robotic capture of non-cooperative target by adaptive extended Kalman filter based visual servo[J]. Acta Astronautica,2016,122:209-218.doi:10.1016/j.actaastro.2016.02.003 [26] XIONG K,WEI C,LIU L. Robust extended Kalman filtering for nonlinear systems with stochastic uncertainties[J]. IEEE Transactions on System Man & Cybernetics,Part A,2010,40(2):399-405. -
计量
- 文章访问数:33
- 被引次数:0