【作者】 赵德勇；

【导师】 王正明；

【作者基本信息】 国防科学技术大学 ， 控制科学与工程， 2007， 博士

【摘要】 以卫星轨道动力学信息的稀疏参数建模和测量信息的参数化建模为基础的参数化信息融合数据处理技术是实现天地基综合信息网中卫星联合定轨需要突破的关键技术之一。本文以系统科学理论和系统工程技术为指导,以轨道动力学信息的稀疏参数建模和测量信息的参数化建模为突破口,研究联合定轨参数化信息融合的理论、模型与方法,并在基于双星定位系统的近地卫星联合定轨涉及的几个典型研究层次中获得应用。本文的主要工作和创新点如下:1)联合定轨的需求分析及定轨原理建模。通过对联合定轨系统的功能需求和数据需求分析,建立了以定轨系统结构和测量系统特征分析、轨道动力学信息的稀疏参数建模和测量信息的参数化建模、非线性半参数联合估计算法设计、联合定轨精度综合评估为内涵的通用联合定轨数据处理流程,拓展和完善了传统联合定轨的概念内涵和研究外延,在此基础上建立了基于双星距离和观测数据的联合定轨基本模型及其数值融合算法。2)动力学模型稀疏参数建模及高精度表示。针对卫星轨道动力学模型建模误差的存在,改进了传统的卫星轨道动力学模型误差补偿方法,提出了一种物理模型和基于稀疏参数表示和时间序列分析的数学模型相结合的卫星轨道动力学模型高精度表示方法,在分别建立轨道摄动偏差信号的基于小波分解的加权迭代稀疏参数估计算法和轨道摄动残差信号的时间序列参数估计方法的基础上,建立了轨道动力学高精度表示模型的改进的Gauss—Newton迭代求解算法。理论分析和实验结果表明,该建模方法是对物理参数模型建模误差的一种有效补偿,卫星联合定轨精度得到较大程度的改善。3)测量模型误差的非线性半参数建模及精度评估。在建立多测元观测数据的联合定轨多源融合测量模型及其多结构非线性融合处理算法的基础上,针对联合定轨数据处理中的非线性影响因素导致的测量模型误差,提出了一种参数化建模和非参数分量表示相结合的非线性半参数联合定轨模型建模方法,在此基础上建立了参数估值的偏差修正算法和基于数据融合处理的联合定轨精度综合评估方法。理论分析和实验结果表明,考虑测量模型误差的基于正则矩阵补偿的半参数估计方法及其偏差修正算法能够有效改善卫星的联合定轨精度。4)联合定轨应用技术的分层次研究方法。以基于双星定位系统的近地卫星联合定轨作为工程应用背景,提出了联合定轨应用技术的基于观测数据层、模型结构层、策略融合层的分层次研究方法,分别建立了同质观测数据的二步系统误差消除的方差分量估计加权算法以及异质观测数据的方差分量估计和模型结构分析相结合的综合加权算法、基于加权因子的多结构非线性联合定轨模型及其最优加权算法、基于动力学和运动学定轨策略融合的一体化联合定轨模型及其加权融合估计算法,三个层次的最优加权估计算法能够进一步抑制轨道动力学建模误差、测量模型误差等非线性影响因素对联合定轨精度的影响,最终的卫星轨道参数估计精度得到了进一步的改善。更多还原

【Abstract】 The parameterized information fusion technology based on sparse parameter modeling of satellite trajectory dynamic information and parameterized modeling of measurement information is the key technology of satellites combined orbit determination(COD) in space-ground-based synthetical information network. Taking the system science theories and system engineering technologies as guidances, the theories, models and methods of COD parameterized information fusion technology are studied in the thesis taking sparse parameter modeling and parameterized information fusion as breakthrough point, and the applied technology of low earth orbit(LEO) satellites COD based on bi-satellite positioning system (BPS) is discussed in representative research hiberarchy.The contributions and innovation points of the thesis are listed as follows.1) Requirements analyses and fundamental modeling of COD. The general COD data processing course which include COD system structure and measurement system characteristic analyses, sparse parameter modeling of trajectory dynamic information and parameterized modeling of measurement information, nonlinear semiparametric combined estimation algorithm design, COD precision synthetical evaluation is constructed by analyzing of COD system function requirements and data requirements, which expands and perfects the conception connotation and research denotation of traditional COD. Based on these, COD basal mathematic model based on bi-satellite range sum data and its numerical fusion algorithm are constructed.2) Sparse parameter modeling of orbit dynamics model and high precision denotation. Aiming at modeling errors of satellite dynamic models, a high precision denotation method of satellite orbit dynamics based on physics model and mathematics model with sparse parameters representation and time series analysis is presented by improving traditional method of dynamic model error compensation. Based on upbuilding weighting iterative sparse parameters estimation algorithm of trajectory perturbation deviation signal based on wavelet decomposition and time series parameters estimation algorithm of trajectory perturbation residual error signal, an improved Gauss-Newton iterative algorithm of high precision denotation model is brought forward. Theoretical analyses and simulation experiments results show that the modeling method can achieve effective compensation for physics model errors, and the COD precision can be gained better improvement.3 )The model error modeling method based on nonlinear semiparametric regression and precision estimation. Based on building COD multi-source fusion measurement models and multi-structural nonlinear algorithm, a modeling method of nonlinear semi-parametric COD model which associated parameterized modeling with non-parametric components expression is put forward aiming at model errors of COD nonlinear influencing factors, and the deviation correction simulation algorithm and COD precision estimation method based on data fusion process are built. Theoretical analyses and simulation experiments results show that the semi-parametric estimation method based on regularization matrix compensation and its deviation correction algorithm can ameliorate precision effectively considering model errors modeling.4) The layered research method of COD applied technology. Taking LEOs COD based on BPS as the engineering application background, a layered research method of COD applied technology based on observation data layer, model structure layer and strategy fusion layer is put forward. The improved variance component estimation optimal weighting method of homogeneous data with two-step system errors correction and the integrated weighting method based on model structure characteristics analyses and VCE estimation of heterogeneous data, multi-structural nonlinear regressive model and optimal weighting algorithm of COD system based on weighting factor, integrative COD model based on dynamic and kinematic orbit determination strategy fusion and the fusion weighting algorithm are constructed, and theoretical analyses and simulation computation results show that optimal weighting estimation algorithm of three layers can restrain nonlinear influence factor including dynamic modeling error and measure model errors to the effects of COD precision.更多还原