【作者】 涂佳；

【导师】 吴翊； 易东云；

【作者基本信息】 国防科学技术大学 ， 数学， 2012， 博士

【摘要】 分布式InSAR卫星系统是将卫星编队技术与干涉合成孔径雷达（InterferometrySynthetic Aperture Radar，InSAR）技术相结合的新概念新体制雷达系统，极大地拓展了SAR卫星系统的总体性能，具有广泛的发展前景。然而，该系统的实现在基础理论和技术层面上面临着许多挑战，星间基线的高精度确定就是其中之一。鉴于此，本文以分布式InSAR卫星系统为背景，利用星载双频全球定位系统（GlobalPositioning System，GPS）作为测量手段，根据分布式InSAR任务尤其是星间基线确定与单星精密定轨、星间高精度相对定位之间的关系，围绕“单星—双星—基线”这一主线，开展了分布式InSAR卫星系统的高精度星间基线确定研究。本文的研究工作和贡献主要体现在以下三个方面：针对单颗卫星绝对位置、速度的确定问题，详细地研究了基于双频GPS的单星精密定轨方法，提出了一种综合考虑GPS接收天线相位中心变化系统误差和随机误差的混合误差建模与修正方法，有效地提高了低轨卫星的精密定轨精度。首先，探讨了星载双频GPS数据预处理的方法，研究了基于双频GPS的单星简化动力学精密定轨方法；其次，在GPS载波相位观测模型和GPS接收天线相位中心模型的基础上，提出了一种综合考虑GPS接收天线相位中心变化系统误差和随机误差的混合误差建模与修正方法，该方法按照GPS信号的接收方向，将相位中心变化误差按照系统部分和随机部分分别进行建模，并通过接收方向的相位定轨后验拟合残差的均值和标准差分别对相位中心变化误差混合模型的系统部分和随机部分的标准差进行估计，利用该方法对重力反演与气候实验（Gravity Recovery AndClimate Experiment，GRACE）双星编队实测GPS观测数据进行处理，生成了三种不同类型的轨道解，通过三种不同校验方式表明：经混合误差模型修正后得到的精密轨道解均优于其它两种类型的轨道解。与此同时，通过将由相位中心变化误差均值修正得到的轨道解与由混合误差模型修正得到的轨道解进行轨道比对的结果表明：对混合误差模型的随机部分的修正对于精密定轨而言是不可忽略的。针对双星相对位置、速度的确定问题，系统地研究了基于宽窄巷双差整周模糊度确定策略的高精度星间相对定位方法，提出了一种基于先验相对轨道和钟差解的双差整周模糊度确定方法，有效的提高了双差整周模糊度确定的成功率和相对定位精度，成功实现了GRACE卫星编队1mm星间相对定位。首先，针对参考GPS卫星频繁更换的问题，提出了一种模糊度分段解算的策略，确保在每个分段区间内共视的GPS卫星不发生变化，在此基础上，系统地研究了基于宽窄巷双差整周模糊度确定策略的简化动力学相对定位方法，并利用该方法对GRACE卫星编队实测数据进行了相对定位实验，实验结果表明：双差整周模糊度确定的成功率为84.73%，相对定位的K/Ka波段测距（K/Ka-band Ranging，KBR）系统校验标准差为1.26mm，从而验证了自编算法的有效性和正确性；其次，针对宽窄巷双差整周模糊度确定策略中存在双差宽巷整周模糊度的确定容易受到伪码观测质量影响的问题，提出了一种基于先验相对轨道和钟差解的双差整周模糊度确定方法，该方法首先利用简化动力学单差无电离层组合批处理最小二乘相对定位方法求解编队卫星之间的相对位置解和接收机之间的相对钟差解，并以此作为先验解，然后通过伪码、相位最优加权求解宽巷模糊度，通过对GRACE卫星编队实测数据的处理结果表明：第一，双差整周模糊度确定成功率为89.89%，提高了5%，该处理策略可以有效地克服宽窄巷双差整周模糊度确定策略中存在的问题；第二，相对定位的KBR校验标准差为1.01mm，相对定位的精度得到有效提高，成功实现了1mm星间相对定位；第三，将两种处理策略得到的相对定位解在径向、横向和法向上的比对结果分别为0.43mm，1.01mm，0.81mm，三维结果为1.37mm，该结果反映了新提出的处理策略引起的相对定位解的变化。在基于双频GPS的星间基线测量方案的基础上，详细研究了空间域基线确定中的各个误差项的性质，理论分析了各个误差项对空间域基线确定的影响，提出了一种空间域基线确定的误差建模与仿真分析方法，通过建立了各个误差项的数学模型，仿真分析了单个误差项、部分误差源以及整体误差源对星间基线确定的影响。首先，根据空间域基线的确定原理，将空间域基线确定的误差源划分为与GPS测量有关的误差源和与部位修正有关的误差源两大类，并在此基础上，对两大类误差源中所包含的各项误差的种类及特性进行了分析，理论分析了各误差项对空间域基线确定的影响；其次，根据各项误差的特性，提出了一种空间域基线确定的误差源建模与仿真分析方法，该方法通过对各误差项分别进行建模，利用仿真实验的方式，分析了单个误差项、部分误差源及整体误差源对空间域基线确定的影响，仿真结果表明：第一，与GPS测量有关的误差源是影响空间域基线确定精度的主要因素；第二，GPS相位观测噪声和GPS接收天线安装位置的地面标校误差是与GPS测量有关的误差源中影响最大的两个因素；第三，在地固坐标系中，整体误差源对空间域基线确定的影响为x方向0.500mm，y方向0.500mm，z方向0.452mm，三维影响为0.845mm，可以实现1mm（每轴）精度的分布式InSAR卫星系统星间基线确定。更多还原

【Abstract】 The distributed InSAR satellite system is a new kind of the conceptual radarsystem, which combines the technology of the satellite formation and the InterferometrySynthetic Aperture Radar (InSAR). The entire performance of the SAR satellite systemcan be well enhanced, and the distributed InSAR satellite system has a broad prospect.However, there also exist some challenges in basic theory and technique. The highprecision determination of inter-satellite baseline is just one of challenges. As a result,in this paper, the high precision determination of inter-satellite baseline of thedistributed InSAR satellite system is selected as the research background, and thespacrborne dual-frequency GPS is used for measuring. According to the relationsbetween the distributed InSAR mission, especially the inter-satellite baselinedetermination and precise orbit determination of single satellite, high precision relativepositioning of inter-satellite, the high precision determination of inter-satellite baselineof the distributed InSAR satellite system is studied surrounding the main line of singlesatellite, inter-satellite and baseline.The main work includes three parts:Aiming at the single satellite absolute position and velocity determination, theprecise orbit determination approach of single satellite based on dual-frequencyGPS is studied, and a mixed error modeling and revising approach of the GPSreceiver phase center variation synthetically considered the systematic error andrandom error is proposed, which improves the precision of the precise orbitdetermination of single satellite. At first, the preprocessing of dual-frequency GPSobservations is studied, and the reduced dynamic prcise orbit determination approach ofsingle satellite is analyzed. Secondly, according to the GPS carrier phase observationmodel and GPS receiver phase center model, a mixed error modeling and revisingapproach of the GPS receiver phase center variation synthetically considered systematicerror and random error is proposed. In this approach, the phase center variation error ismodeled as a systematic and a random component each depending on the direction ofGPS signal reception, and the systematic part and the standard deviation of the randompart in the phase center variation error model are respectively estimated by the bin-wisemean and standard deviation values of phase post-fit residuals computed by orbitdetermination.. The GPS observation data of Gravity Recovery And ClimateExperiment (GRACE) satellites are processed, and there types of orbit solutions areproduced. These orbit solutions are tested by three validation methods. The validationresults show that the orbit solutions obtained by the mixed error modeling and revisingapproach are all the best of the three types of the orbit solutions. Meanwhile, by theresults of the orbit comparisions between the orbit solutions obtained by mean value revision of phase post-fit residuals and the mixed error modeling and revising approach,it is shown that the impact of random part of the mixed error model on precise orbitdetermination can not be neglected.Aiming at the inter-satellite relative position and velocity determination, thehigh precision inter-satellite relative positioning approach based on the wide andnarrow lane double difference (DD) integer ambiguity resolution strategy isstudied, and a new DD integer ambiguity resolution strategy based on the a-priorirelative orbit and clock solutions is proposed, the success rate of the DD integerambiguity resolution and the precision of relative positioning are both improved.1mm level relative positioning for GRACE formation can be realized. At first, aimingat the problem that the reference GPS satellite is continually replaced, a partitionstrategy of ambiguity resolution is proposed, which keeps the common reference GPSsatellite unchangeable in every partition interval. Based on this strategy, the reduceddynamic relative positioning approach based on the DD integer ambiguity resolutionstrategy of the wide and narrow lane strategy is studied, which is used for the relativepositioning expriment of GRACE satellites. In this experiment, the sucess rate of theDD integer ambiguity resolution is84.73%, and the standard deviation of K-bandRanging (KBR) system for relative position validation is1.26mm. These resultsillustrate the validation and correctness of the programme. Then, aiming at the problemof the forementioned strategy that the resolution of the DD integer wide-lane ambiguityis easily affected by the quality of GPS P-code observation, a new DD integerambiguity resolution strategy based on the a-priori relative orbit and clock solutions isproposed. In this strategy, the a-priori relative orbit and clock solutions are firstlygenerated by the single difference reduced dynamic ionosphere-free combination batchleast-squares relative positioning approach, and then the wide lane ambiguities aresolved by the optimization of the P-code and phase data. The GPS observation data ofGRACE satellites are processed. The sucess rate of the DD integer ambiguity resolutionis89.89%, and is improved by5%, which shows that this stretegy could overcome theproblem of the wide and narrow lane strategy. The standard deviation of KBR systemfor relative position validation is1.01mm, and the accuracy of relative positioning isalso improved.1mm level relative positioning can be realized. The relative positionsolutions obtained by two strategies are compared. The comparison results in radial,transverse and normal directions are0.43mm,1.01mm and0.81mm repectively. The3-dimensional result is1.37mm. These comparison results also indicate that thechanges induced by the newly proposed strategy.Aiming at the inter-satellite baseline measurement scheme based ondual-frequency GPS, the error characters of the spatial baseline determination arestudied in detail, and the impact of each error on the spatial baselinedetermination is theoretically analyzed. An error modeling and simulating approach for the spatial baseline determination is proposed, in which themathematical model of each error is set up, and the impacts of each error, thepartial error source, and entire error sources on inter-satellite baselinedetermination are analyzed by the the simulations. At first, according to the theoryof the spatial baseline determination, the error sources of the spatial baselinedetermination are classified into two parts, the error source related to GPS measurementand the error source related to baseline transformaion. The classifications and charactersof errors are analyzed, and the impacts of the error items on spatial baselinedetermination are theoretically analyzed. Then, according to the characters of erroritems, an error modeling and simulating approach for the spatial baseline determinationis proposed. In this approach, the error items are modeled separately, and the impacts ofa single error item, the partial error source and the entire error sources on spatialbaseline determination are analyzed by simulations. It is shown by the simulations thatthe error source related to GPS measurement is the main errors for the spatial baselinedetermination, and the carrier phase noise of GPS observation and the fixing error ofGPS receiver antenna are the main factors of the error source related to GPSmeasurement, and the impact of the entire error sources on spatial baselinedetermination in International Terrestrial Reference Frame (ITRF) is0.500mm ofx-axis,0.500mm of y-axis,0.452mm of z-axis, and0.845mm of3dimensions.Therefore,1mm (each axis) level InSAR spatial baseline determination can be realized.更多还原