QQ客服 官方微博 反馈留言

题名	数据质量控制理论及在GPS数据处理中的应
作者	王爱生
学位类别	博士
答辩日期	2007-06
授予单位	中国科学院测量与地球物理研究所
授予地点	武汉
导师	欧吉坤
关键词	全球导航卫星系统 GNSS 全球定位系统 GPS 粗差 粗差转移 周跳 多路径 高阶差分 连续性诊断 疑似粗差 拟准检定法 QUAD 分群准则 平差模型 平滑参数 双低通滤波 特征子序列 系统误差 相关分析 相关观测 虚拟观测 不适定问题 正则化解法 选权拟合 FMSPW 约束方程 质量控制
学位专业	大地测量学与测量工程
中文摘要	测量的数据处理通常使用最小二乘法，因为当观测误差属正态分布时最小二乘估计是无偏估计且方差最小。但是，最小二乘法抵御非偶然误差的能力很弱，因此，要对观测数据进行质量控制，以便采取适当的措施消除这些非偶然误差对观测结果的影响。传统测量由于数据量小，相关性不强，测量数据质量易于控制，而在信息化时代，新的测量技术不断出现，数据以前所未有的速度被采集和更新，因此，研究更加可靠快速的数据质量控制技术显得尤为重要。 GPS是一个基于卫星的导航系统，观测值中的粗差、周跳、多路径误差等极大地制约着GPS定位的精度和可靠性，因此本文在研究数据质量控制的基本理论和方法的基础上，进一步研究如何将这些数据质量的理论用于GPS的质量控制。本文以选权拟合法为主要研究对象，目的是将选权拟合法用于周跳、粗差的探测和系统误差的反演。因此本文推导了选权拟合法的性质，给出了选权拟合法的一种等价模型，说明了拟准检定法是选权拟合法的一种特殊形式；论述了将选权拟合法的理论用于GPS的数据质量控制的具体思路和算法，其一是论述了拟准检定法的分群标准和单位权中误差的选取方法，提出了一种实用的算法——R+1法；其二是将选权拟合法用于均值漂移模型中解算粗差和GPS周跳；其三是将选权拟合法用于半参数模型中来反演系统误差，从而反演多路径误差。针对GPS数据连续性的特点，本文在研究GPS周跳探测时还讨论了基于连续性诊断技术的周跳探测方法，包括利用周跳在高阶差分的数字特征探测和固定周跳的方法、无距离组合联合MW组合探测周跳的方法以及使用双低通滤波器探测周跳的方法。另外在讨论多路径误差时还提供了一种基于相关性分析技术的多路径误差的诊断方法。本文的特色和创新体现在以下内容中： 1）对两种基于连续性诊断技术的周跳探测方案进行了改进，并提出一种新的算法发现了周跳在高阶差分中具有叠加的性质，推导了周跳在高阶差分中的特征子序列；提出了利用周跳在高阶差分中的时序特征探测和固定周跳的方法；算例表明，高阶差分法除了可以探测单个周跳，还可以探测多周跳；提出使用双低通滤波器探测周跳的方法；算例表明，使用双低通滤波器探测单个周跳的精度高于高阶差分法；对无几何距离组合联合MW组合探测周跳的算法进行了改进，改进体现在三个方面，一是将高阶差分法用于探测无几何距离组合和MW组合的周跳，二是将低通滤波用于伪距窄巷组合的平滑，三是对解算得到的L1和L2的周跳浮点解进行条件搜索，将它们固定为整数；算例表明，MW组合和无距离组合联合起来能够用于静态和动态的周跳探测。 2）拓展了选权拟合法的相关理论，开发了选权拟合法新的应用领域比较了正则化算法与选权拟合法的不同，给出了正则化解和选权拟合估计的性质；在广义平差原理基础上推导出了选权拟合法的等价模型和精度计算公式；证明了在选权拟合法模型中，平滑参数其实就是被约束参数的权；给出了选权拟合法中构造正则化矩阵（约束方程）的差分公式；在选权拟合法原理下推出了相关观测条件下拟准检定法的真误差计算公式；将选权拟合法用于系统误差的反演，并进一步用于多路径误差的反演；将选权拟合法用于均值漂移模型中进行抗差估计。 3）深入探讨了拟准检定法的有关理论，提出了用拟准检定法探测周跳的实用算法以选权拟合法的原理为基础推出了相关观测条件下拟准检定法的真误差计算公式；提出拟准检定法的新的分群准则；提出在拟准检定时采用先验的中误差作为粗差标准参数；利用拟准检定法可以一次探测多个粗差的特点，将拟准检定法用于周跳探测，算例表明，拟准检定法不仅在载波相位非差模型效果很好，在双差模型上的效果同样很好；提出了一种实用的探测周跳的算法——R+1法。 4）提出了诊断多路径误差的新思路，并利用选权拟合法反演多路径误差提出将相关性分析技术用于辨识多路径误差的方法；使用选权拟合法反演系统误差的算法反演多路径误差，经过对连续两天的GPS基线向量的系统误差的计算结果表明，这两天的系统误差呈现明显的重复性，说明系统误差主要是由于多路径误差引起的，在计算时，使用了系统误差的三阶差分等于零的约束条件，说明多路径误差可以用二阶多项式表示
英文摘要	Least Squares (LS) is the most popular techniques in surveying data precessing, since when observation errors conform the normal distribution the LS estimates are unbiased and have the minimum variances. Unfortunately, the least square method has weak ability to defense non-accident errors, thus we must carry on the quality control of observations and take measure to remove the influences of non-accident errors on the adjusted results. Due to less quantity and correlation in tradition surveying, it is easy to perform the quality control. But in the informationization ear many new observation techniques appear, the surveying data are rapidly collected and updated as never expected. It is worthwhile to develop a much efficient and reliable quality control method to deal with the large quantity of data. GPS is a satellite-based navigation system, the gross errors and cycle-slip and multipath errors greatly influenced the positioning accuracy and reliability. Thus, after being introduced and researched, the quality control theory and methods were further discussed as how to be applied to GPS data processing. The fitting method by selection of parameter weights (FMSPW) is the research subject in this dissertation. The goal is how to apply this theory to the detection of cycle-slip, gross errors and systematical errors. Thus the property of FMSPW are deducted, a type of equivalent mathematics model is given; it is proved that the method of quasi-accurate detection of gross error (QUAD) is an especial form of FMSPW. Then the detailed plan and algorithm of the application of FMSPW on GPS data quality control is described, one of which is the application of QUAD on GPS cycle slips detection, including the group standard and the selection of mean square error of unit weight in QUAD. Proposed a practical method for detecting the cycle slips in GPS carrier phase which is called R+1; second of which is the application of the FMSPW in the expectation drift model to resolve outlier and GPS cycle slips; third of which is the application of FMSPW in semi-parametric model to take inversion of systematic errors, and further recover the multipath errors. According to the continuous of GPS data, when cycle slips was discussed in this paper the techniques of cycle slip detection based on continuity diagnosis are studied, including the higher order differential method, combined method by geometry-free combination and MW combinations, and double low pass filters. By the way when the multipath effect was discussed, an identify method of multipath effect based on correction analysis was proposed. The characteristics and innovations in this dissertation are as follows: 1) Making the improvement to two kinds of cycle slips detecting plans based on the continuous diagnosis technology and proposing a new algorithm for cycle slips detecting It is discovered that cycle slips has additivity in the higher order differential, the added characteristic subsequence are derived; the method of detecting and fixing cycle slips using the characteristic high order differential subsequence about cycle slips is proposed; The numerical example indicated that this method suitable to the single slip and many slips; the method of detecting cycle slips using the double low pass filter is proposed; The numerical example indicated that this the precision of detected cycle slips using the latter are higher than using the former; the improvement to the previous method of detecting cycle slips using combination of the MW combination and geometry-free combination is made, which include three aspects; (i) Using least square estimator to locate and fix cycle slips according to whose characteristic in highly order differential time series； (ii) Using double low pass filter to smooth code pseudoranges；(iii) Finding integer number of cycle slips from float value of cycle slips using search algorithm under certain conditions．The numerical example indicated that this method can be used to detect cycle slips in the either static GPS or dynamic GPS. 2) Making the development of the theories about FMSPW, exploring new application domain of FMSPW The differences in principle between the regularization algorithm and FMSPW are demonstrated, a few properties of the regularization algorithm and FMSPW are given; a kind of equivalent mathematic model to FMSPW and its precision formula are derived based on the generalized adjustment principle; it is proven that the smoothing parameter in this equivalent model is actually the weight of the constrained parameters; the general formula of the regularization matrix (constraint equation) of FMSPW is given; the estimation formula of the true errors in the method of quasi-accurate detection (QUAD) is derived under the correlated observation condition from the principle of FMSPW; FMSPW is used to make inversion of systematic errors and further make inversion of the multipath error; FMSPW is used to detect gross error in the expectation drift mode. 3) Probing deeply into the related theory of QUAD, proposing a practical algorithm of detecting cycle slips with QUAD The estimation formula of the true errors in the method of quasi-accurate detection (QUAD) is derived under the correlated observation condition from the principle of FMSPW; the group standard of QUAD is improved. It is proposed that distinguishing gross error should reference to prior unit weight standard deviation; due to the property that the QUAD can detect multiple blunders in observations, it can be used to detect cycle slips in GPS carrier phase observations, numerical example indicated that the QUAD can detect cycle slips in GPS single difference model of relative positioning as effectively as in GPS double difference model , one kind of practical algorithm of detecting clips using QUAD called R+1 method is proposed 4) Proposing a new idea of diagnoses multipath error and a new method by using FMSPW to inverse parameters of multipath error It is proposed that the multipath error can be recognized using the correlation analysis technique and inverted using FMSPW, through the correlation analysis to time series of GPS baseline of identical time interval of two adjacent days we can decide whether the two sequences have a very significant repeatability, which indicate that the systematic errors are caused by the multipath errors or not; in the computation, the condition must enforced that the third order differential of systematic errors equals to zero, which implies that the multipath error might be expressed into the second-order polynomial about time
语种	中文
公开日期	2013-01-17
内容类型	学位论文
源URL	[http://ir.whigg.ac.cn//handle/342008/3686]
专题	测量与地球物理研究所_学生论文_学位论文
推荐引用方式 GB/T 7714	王爱生. 数据质量控制理论及在GPS数据处理中的应[D]. 武汉. 中国科学院测量与地球物理研究所. 2007.

个性服务

查看访问统计

相关权益政策

暂无数据

收藏/分享

除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。