PDF《GPS 精密定位定轨后处理 算法与实现》

分类号 P228 密级 U D C 编号

10486 武汉大学博士学位论文GPS 精密定位定轨后处理 算法与实现 研究生姓名: 姚宜斌指导教师姓名、职称: 刘经南教授 陶本藻教授 学科、专业名称: 大地测量学与测量工程 研究方向: 测量数据处理理论与方法 二零零四年四月二十日 Research on the Algorithm and Realization of Post-processing for GPS Precise Positioning and Orbit Determination By YAO YIBIN Supervised by Prof.

Liu Jingnan &

Prof. Tao Benzao Wuhan University School of Geodesy and Geomatics April.

2004 郑重声明 本人的学位论文是在导师指导下独立撰写并完成的,学位论文没有剽窃、抄袭、 造假等违反学术道德、学术规范和侵权行为,本人愿意承担由此而产生的法律责任 和法律后果,特此郑重声明. 学位论文作者(签名) :

2004 年4月20 日・i・ 摘要卫星精密定轨是卫星对地观测技术发展和应用的基础,没有先进的定轨技术和 相应的软件保证获得高精度轨道,就不可能进行精密定位和开拓其它领域的应用. 我国卫星定位技术的应用已有

20 多年的历史,但是目前卫星精密定轨、定位软件还 主要依靠引进的软件,这种依赖国外软件的状况很难取得突破性的创新成果.只依 靠国外软件已无法满足我国快速发展的卫星计划的实际需要,也不利于国防安全的 保障,所以应尽快开发具有我国特色和自主知识产权的卫星精密定位定轨软件系统. 卫星精密定位定轨研究实际包含两部分的内容,一部分是利用原始观测数据进 行地面站间基线单天解和单天轨道确定;

另一部分是后处理部分,即利用单天解的 轨道和基线处理结果进行多种参数联合估计和轨道综合.高精度定位定轨后处理的 数据源包括轨道参数、各类地球动力学模型参数、站坐标参数、站速度参数、对流 层参数、整周模糊度参数乃至卫星和接收机参数等,各类参数具有复杂的函数相关 性和误差相关性,处理起来比较复杂. 本文研究的内容主要涉及卫星精密定位定轨研究的后处理部分,主要是研究利 用单天解的轨道和基线处理结果进行多类参数(测站坐标参数、测站速度参数、地 球自转参数、定轨参数等)的联合估计和轨道综合问题,在此领域国内尚没有开展 系统性的研究.本论文研究的成果和内容具体包括: (1) 提出利用 SINEX(Solution INdependent EXchange Format)文件作为 GPS 精 密定位定轨后处理的主要输入文件,讨论了 SINEX 文件结构及其与法方程的相互转 换问题.SINEX 文件作为数据解文件的标准格式,包含 GPS、VLBI、SLR 等所提供的 基线解算或者是网平差后的点位坐标、点位协方差、点位先验坐标、点位先验方差 信息,可以通过 SINEX 文件恢复法方程,通过法方程叠加实现多期、多类、多子网 数据的联合处理. (2) 在分析基于基线模式的常用的 GPS 网平差模型的基础上, 提出直接以所涉及 的测站坐标信息、地球自转参数信息等为观测量的基于坐标模式的广义网平差模型. 分析了基于基线模式和基于坐标模式的模型之间的相互联系和区别,总结了基于坐 ・ii・ 标模式的广义网平差模型在处理多类、多期参数中的优越性,概括了其处理的通用 方法.提出采用基于等价方差-协方差的稳健最小二乘估计方法解决基于坐标模式下 的广义网平差中的粗差问题. (3) 从随机模型误差和函数模型误差两个方面,对参数估计中先验约束的影响进 行了详细的分析,推导了先验约束值的误差与先验约束方差的误差对参数估计的影 响,提出先验约束的消除是附加约束的逆过程,给出了先验约束消除的算法. (4) 以附加约束条件的间接平差模型为例,推导了附加额外约束条件对平差结果 的影响,并以平差后的单位权中误差减小为条件,分析了附加额外约束产生增益的 条件,说明附加额外约束条件并不都能增强解的性状. (5) 提出利用参数转换的方法对参数先验初值的统

一、参数合并、法方程叠加、 新参数的引入、法方程的标准化等基本算法进行统一,对基于坐标模式的广义网平 差模型的算法实现所涉及的若干关键问题进行了详细推导,为基于坐标模式的广义 网平差的实现提供了理论基础. (6) 重点研究了基于坐标模式的测站坐标参数、测站速度参数、地球自转参数的 估计问题,并编制了相应的程序.通过算例证明了目前在测站速度的估计中将测站 坐标变化假释为匀速运动与客观事实存在差异.对于基于周解的测站坐标序列,利 用回归分析的方法建立了测站速度变化经验模型.提出对于 IGS 跟踪站的速度值, 单单用一个线性的量去估计它存在偏差,应考虑建立周期性变化模型,或考虑采用 分段线性化的方法,每隔一定的时间给出一个对应的速度. (7) 采用轨道叠加的方法,对利用单天解轨道合成多天解轨道算法进行了详细推 导,并通过算例证明了轨道综合对提高轨道产品精度的有效性.特别是在局域定轨 中,轨道综合对提高定轨精度的贡献尤其明显.同时也对不同分析中心所处理的多 天解轨道的综合合成进行了详细分析推导,并给出了笔者处理的结果与 IGS 分析中 心结果的比较. 【关键词】 : GPS 定位定轨;

坐标模式;

广义网平差;

参数估计;

轨道综合 ・iii・ ABSTRACT Satellite precise orbit determination(POD) play a very important role in Earth Observing System (EOS). Precise positioning can'

t be done without advanced POD technique and software, which ensure to achieve the accurate orbit of satellite. In the past twenty years, the technology of satellite positioning has been widely developed and applied in china. But the software of precise orbit determination and precise positioning we used always comes from foreign country. Such situation will impede us to make outstanding and creative achievement. Just relying on the imported software can'

t meet the practical need of developing the satellite plan in our country, and also make against national defense safety guarantee. In a word, we should develop the satellite precise orbit determination and precise positioning software with our own property and characteristic as soon as possible. Research on satellite precise positioning and orbit determination includes two main parts: The first part is to process the raw observation data of tracking stations on the earth to get the baseline single day solution and single day orbit determination. The second one is post-processing part, which use the single day orbit resolutions and baseline solutions to estimate the parameters and integrate orbit. The parameters to be estimated in the post-processing include orbit parameters, geodynamic model parameters, station coordinate and velocity parameters, tropospheric parameters, ambiguity parameters, satellite and receiver clock parameters and so on. Complicated functional correlation and error correlation between these parameters cause complex post-processing. The post-processed part of precise position and precise orbit determination is mainly studied in the dissertation. Using the single day solution orbit and the single day baseline solution to estimate various parameters such as station coordinate parameters, station velocity parameters, earth rotation parameters, orbit parameters and to integrate the satellite orbit is investigated in detail. Such study has not been developed systematically yet in our country before. The main achievements and concerned items of the dissertation are as following: ⑴ Using SINEX(Solution INdependent EXchange Format) files as the main input files in post-processing of GPS POD and precise positioning is proposed in the dissertation. The SINEX file structure and transformation between SINEX file and the normal equation are discussed. SINEX file is the standard format of solution file, baseline solutions derived from GPS, VLBI, SLR etc. and points coordinates, points covariance, ・iv・ apriori coordinates and apriori variance information can be found in SINEX file. Data processing combined multi-stages, multi-types and multi-subnets can be done by adding normal equations recovered from SINEX files. ⑵ Based on the analysis of traditional GPS network adjustment model with baseline pattern, a new generalized network adjustment model with coordinate pattern is brought forward. station coordinates, earth rotation parameters (ERP) etc. are used directly as observations in the new model. The relationship and the different of the two models are analyzed. The advantages of the generalized network adjustment model with coordinate pattern in parameters estimation of multi-types and multi-stages are summarized. The common ways of dada processing of the new model are generalized. The robustified least squares estimator on equivalent variance-covariance method is used to solve the outliers problem on the generalized network adjustment model with coordinate pattern. ⑶ The influence of the apriori constraint in parameter estimation are analyzed in detail with consideration of function model error and stochastic model error. The influence of the error of the apriori constraint value and the apriori constraint covariance on estimated parameters is derived. This study also demonstrates that the elimination of the apriori constraint is the inverse process of additional constraint. The algorithm of elimination the apriori constraint is presented. ⑷ Based on indirect adjustment model with additional constraint condition, the influence of additional extra constraint conditions on adjustment results is deduced. With the condition of decreasing the variance of unit weight (variance factor)

0 ? ? , the condition of that adding extra constraint brings plus is analyzed. The conclusion is that the quality of solution can'

t be improved with additional constraints all the time. ⑸ Using the way of parameters transformation to unify the basic algorithms such as unification of the apriori initial value of parameters, the parameters merging, normal equations adding, unknown parameters introducing, normal equation normalization and so on is proposed. Some key issues in the algorithms realization of generalized network adjustment model with coordinate pattern are deduced in details, which is the theoretic base used to realize generalized net adjustment model with coordinate pattern. ⑹ The parameters estimation issues of the new model have been mainly investigated. The estimated parameters include stations coordinate, station velocity, earth rotation and so on. In addition, the software has been developed in the research. The examples demonstrate that there is difference between the true fact and estimated velocity if we ・v・ assume the station velocity keep constant when we estimate the stations velocity. Moreover, the empirical model of the velocity variation of the station has been established by regressive analysis method based on week solution coordinate series of the station. With respect to the velocity of IGS tracking station, it had better to model the variation periodically or to give a velocity periodically using piece-wise linear function rather than using a linear variable to estimate its bias. ⑺ The algorithm of using single day solution to integrate multi-day solution by the way of orbit overlaying are conducted in details. An example is given, which proves that the orbit integration method can improve the accuracy of orbit efficiently. Especia........