节点文献
基于天基测角信息的空间非合作目标跟踪算法及相关技术研究
Research on Non-cooperative Target Tracking Algorithms and Related Technologies Using Space-based Bearings-only Measurements
【作者】 刘光明;
【导师】 廖瑛;
【作者基本信息】 国防科学技术大学 , 航空宇航科学与技术, 2011, 博士
【摘要】 空间目标监视是国家空间战略信息获取的重要途径,而基于天基测角信息的空间非合作目标跟踪定轨技术是实现我国天基空间目标监视系统亟需突破的关键技术。论文以天基空间目标监视系统为主要研究背景,以参数估计理论、连续同伦方法、非线性系统可观性理论、非线性滤波理论、联邦滤波理论及抗差估计理论为理论基础,重点研究了基于天基测角信息的初轨确定、跟踪滤波、联合定轨、抗差自适应跟踪定轨及目标编目维护等关键技术。本文采用仿真实验与数学推理相结合的研究思路,保证了模型和算法的正确性,初步解决了空间目标天基仅测角跟踪中的若干关键问题。论文的主要研究成果如下:研究了基于天基测角信息的短弧初轨确定方法。(1)提出了短弧初轨计算中两种迭代初值选取方法:考虑J 2项摄动的简单迭代法和基于经典高斯法的信赖域法。(2)提出了基于连续同伦算法的天基仅测角短弧初轨确定方法,该方法在Laplace改进法基础上,采用了考虑轨道摄动影响的单位矢量法,利用连续同伦算法求解天基仅测角条件方程组,实现了单星观测与双星立体观测两种跟踪方式下的初定轨。研究了天基仅测角跟踪系统的可观测性分析方法。(1)依据以微分几何为基础的非线性控制系统可观性理论,结合系统的状态方程和测量方程,提出了天基仅测角跟踪的可观测性分析方法,证明了近圆轨道上运行的单个观测平台对空间目标仅测角跟踪是可观测的。(2)为了描述系统可观测性的强弱程度,引入跟踪滤波误差下限与系统可观测度,结合仿真结果分析了系统可观测度的影响因素。研究了分布式卫星对空间目标的联合定轨技术。(1)提出了基于平方根UKF的编队双星对空间目标的联合定轨方法,将地面站对编队双星的观测数据、星间测量数据及编队双星对空间目标的测角数据融合起来进行联合定轨,整网确定出编队双星和空间目标的状态。(2)提出了基于分布式Sigma-Points信息滤波的广义联邦SPIF算法,将Carlson联邦滤波扩展到主滤波器与局部滤波器维数不同的情况,并采用“条件重置”原则进行状态重置。仿真实验表明,该算法可以提高多滤波器系统的数据处理效率和容错性,可应用于多星对空间目标的联合定轨。研究了基于抗差自适应滤波的跟踪滤波算法。(1)针对状态方程扰动和观测异常会严重影响EKF滤波性能的问题,提出了抗差自适应EKF算法,采用将观测噪声协方差阵在线估计和自适应因子相结合的方式,从整体上平衡观测数据和状态预报对系统状态估计的贡献。(2)针对实际的跟踪滤波过程中噪声统计特性时变会导致传统UKF滤波精度下降甚至发散的问题,提出了基于极大验后估计的抗差自适应UKF算法,研究将时变噪声统计特性在线估计和抗差自适应因子相结合,有效抑制了粗差的影响,提高了滤波精度和稳定性;并参考传统UKF稳定性分析方法和Cramer-Rao下限理论,以数学推理方式对该算法进行了滤波性能分析。提出了面向天基仅测角跟踪应用的双行轨道根数生成方法。(1)针对空间目标的双行轨道根数(TLE)拟合过程可能出现奇点的问题,在TLE和SGP4模型的基础上,提出了基于无奇异轨道根数的TLE拟合算法。(2)考虑到天基观测平台的数据处理能力有限,通过引入带有自适应遗忘因子的递推最小二乘算法,提出了面向天基仅测角跟踪应用的TLE生成方法。仿真实验表明,该方法迭代收敛速度较快,长弧段轨道预报精度稳定,可应用于近地空间目标的监视跟踪。本文系统研究了基于天基测角信息的空间非合作目标跟踪算法及相关技术,在该领域取得了一些探索性研究成果,为今后进一步开展研究和实验奠定了理论基础,同时也对我国空间目标天基监视技术有一定的参考价值。
【Abstract】 The surveillance of space objects is the important approach to obtain national space stratagem information. The space-based bearings-only tracking and orbit determination technology for non-cooperative space target (NCST) is a critical technology for realizing and establishing our space-based surveillance system. In this dissertation, we stress on researching the initial orbit determination, tracking filter, combined orbit determination, robustly adaptive tracking, and space objects catalog maintenance and so on. We make full use of some mathematical tools including parameter estimation theory, homotopy continuation theory, observability theory of nonlinear control system, nonlinear filter theory, federated filter theory, and robust estimation theory to carry out the research on the key technologies. This dissertation adopts the combination of simulation test verification and mathematics reasoning as the research strategy so as to insure the correctness of models and algorithms researched above. Some key problems have been solved basically for the space-based bearings-only tracking technology of space objects in this dissertation. The main works are as follows:The technologies of short-arc initial orbit determination (IOD) with space-based bearings-only measurements. (1)Two methods of initial iteration values selection in short-arc IOD are proposed that are the simple iteration algorithm considering the J 2 perturbation and the trust region algorithm based on the conventional Gauss method. (2)Based on the improved Laplace method and the unit-vector-method of perturbation orbit, the homotopy continuation theory is introduced into the short-arc IOD method. Then the initial orbit of the NCST with the single satellite observation and two-satellite formation observation can be determined by the presented method.The observability analysis method of the space-based bearings-only tracking system. (1)According to the state equations and measurement equations of the space-based bearings-only tracking system, the observability analysis method is proposed based on the observability theory of nonlinear control system. And the observability of the single satellite with near circle orbit to NCST bearings-only tracking system is proved. (2)The CRLB theory and system observability degree are introduced to describe the relative strength of the system observability, and the influencing factors of the system observability degree are analyzed with simulation results.The technologies of distributed satellites to NCST combined orbit determination. (1)Based on fusion of ground-based observations, inter-satellite measurements and NCST bearings-only measurements, the integrated real-time orbit determination with SR-UKF algorithm is proposed. (2)The generalized federated SPIF algorithm is presented based on distributed sigma-points information filtering, and the method is designed to solve the problem that the master filter dimension does not equal the local filter dimension. Then the information sharing principle with“condition resetting”is used to ameliorate the efficiency of data-processing and fault tolerance.The robustly adaptive filter algorithm of the space-based bearings-only tracking system. (1)Considering the EKF declines in performance when the kinematic state disturbing and measurement outliers,the robustly adaptive EKF algorithm is proposed, and the algorithm combines the noise statistic estimator and adaptive factors to balance the contributions of observation data and predicted state. (2)Aiming at the traditional UKF declines in accuracy and further diverges when the prior noise statistic is unknown and time-varying,the robustly adaptive filter algorithm is proposed based on maximum a posterior estimation. Then the time-varying noise statistic estimator and adaptive robust factors are combined to restrain the influence of abnormal value effectively, on the other hand, enhance the filter precision and stability. And the performance analysis of the presented algorithm is done based on the stability analysis method of traditional UKF and the CRLB theory.The new two-line elements (TLE) fitting method applied to the space-based bearing-only tracking system. (1)Due to the singularity existence in the iterative approximation procedure, the TLE sampling fitting method is put forward according to the non-singular transformation of orbital elements. (2)The new TLE fitting method introduces the adaptive forgetting factor recursive least-squares algorithm to improve the efficiency of data-processing. Numerical simulations indicate that the method can enhance the iterative rapidity of convergence and the accuracy of forecasting orbit, especially for near-earth space objects.This dissertation investigates systemically the space-based bearings-only tracking algorithms and related techniques. The exploratory research achievements of this dissertation would avail to pave the way for further researches and experiments, and also have certain reference values for our space-based surveillance techniques.