基于原始观测值的GNSS统一快速精密数据处理方法

发布时间：2018-07-05 14:27

本文选题：多系统多频率 + 原始观测值　；参考：《武汉大学》2015年博士论文

【摘要】：全球导航卫星系统(Global Navigation Satellite System, GNSS)是二十世纪人类最伟大的科技成果之一,且已经在大地测量、地球动力学以及大气研究等多个领域得到广泛应用。近年来,随着我国北斗导航定位系统(BDS)二代的全面运行、欧洲伽利略系统(GALILEO)的不断发展以及美国GPS系统的现代化,GNSS已经从单系统发展为多系统并存、双频服务发展为至少三个频率服务的局面,这给目前的GNSS数据处理方法带来了极大的挑战。一方面,目前所有公开的GNSS产品均是基于无电离层组合观测值或差分观测值的数据处理方法获得,然而这种方法己经无法完全满足当前多系统多频率环境下的高精度GNSS数据处理要求了。首先为了充分利用各个系统之间的优势,得到更稳定可靠的结果,并为其他用户提供兼容统一的改正数信息,必须将多系统多频率的数据在观测值层面进行统一处理；其次为了尽可能保留原始观测值中的信息,必须在原始观测值而不是差分或组合观测值上建立观测方程。针对此,本文详细研究了一种基于原始观测值的数据处理方法,任意GNSS系统任意频率的数据均能统一采用这种方法进行数据处理。另一方面,计算效率低一直是GNSS领域中未能很好解决的问题。即使仅针对GPS系统,采用组合或差分观测值进行数据处理的方法,几乎目前所有著名GNSS软件都难以或无法同时对含有数百甚至上千个测站的GNSS网进行整网数据处理。然而相对当前组合观测值或差分观测值数据处理方法,原始观测值数据处理方法耗时更为严重,即使针对含有约100个测站的GNSS网,也难以进行日常常规处理。虽然先分子网解算,再合并子网结果的策略可在一定程度缓解这种矛盾,但由于子网问的公共站多次被使用,影响了整网解的严密性。为此,本文提出了一整套能加速GNSS数据处理的策略,不仅可使基于无电离层组合观测值的数据方法可以在短时间内处理含有数百甚至上千测站的GNSS网,还能有效加速原始观测值数据处理。本文的主要工作和贡献为：(1)归纳总结了当前GNSS数据处理方法面临的挑战；系统地研究了基于原始观测值的多系统多频率GNSS数据处理模型；给出了分离钟差参数、系统偏差／频率偏差参数、DCB参数等多类参数的方法；讨论了电离层参数、DCB参数,模糊度之间的相关性；并探讨了在不施加电离层时空约束条件下的模糊度固定方法。(2)利用原始观测值数据处理方法初步探讨了GPS和BDS频率之间的兼容性。实验发现：GPS L5的残差约为L1/L2残差的3倍,可达9～10mm,而且连续数天内其变化量级和变化趋势基本一致,这意味着GPS L5可能含有某种与L1/L2不兼容的系统偏差。而BDS三个频率的残差均在2-3mm,没有发现类似的情况。实验还发现L5码伪距上的DCB估值在数天甚至一个月内都十分稳定,月平均值的标准差优于3cm。(3)评估了在使用两个频率且不施加电离层约束的条件下,基于原始观测值的GNSS数据处理方法与基于无电离层组合观测值的数据处理方法得到的产品的异同。结果发现两种方法计算得到的轨道、钟差、对流层以及系统间偏差均无明显差异,这表明当使用两个频率且不施加电离层时空约束时,基于原始观测值的数据方法与无电离层组合观测值的数据方法基本等价。(4)提出了利用PPP模糊度技术在统一基准下生成Carrier-range的新方法,并在此基础上设计了一整套高效率GNSS数据处理流程。本文从PPP定位的基本观测方程出发,证明了利用单站模糊度技术在统一基准下生成Carrier-range的可行性,并设计了整套高效率GNSS数据流程。主要包括六个步骤：采用全球均匀分布的适合的网计算精密轨道和卫星钟差；利用定轨过程中得到的浮点模糊度计算卫星端的UPD；固定卫星轨道和钟差,逐一测站进行PPP定位,并利用生成的UPD进行单站模糊度固定；然后将载波相位观测值转换成Carrier-range,并将其写入新的RINEX中；重复上两步直至所有测站均生成了新的RINEX文件；利用含有Carrier-range的新RINEX文件进行最终的网解。最终网解不再需要估计或者仅需要估计少量模糊度,其计算效率将大幅度提高。最终解算可采取两种模式,一种仅采用Carrier-range进行计算,此时UPD参数将完全被卫星钟差吸收,不再需要估计,卫星钟差也由于吸收了UPD,成了“整数钟”。另一种方法是同时采用伪距和Carrier-range,此时为了使伪距和Carrier-range钟差定义一致,在Carrier-range上需要估计一组UPD参数,此时钟差绝对基准由伪距决定,与IGS定义保持一致。(5)统计分析了UPD参数的稳定性,提出了UPD参数估计方法。通过对实验数据的分析,发现非经历地影的BLOCK IIA GPS卫星的窄巷UPD在24h内变化平稳,其一天平均值的标准差大都不大于0.05周,而经历地影的BLOCK IIA卫星在出地影后,窄巷UPD可能发生0.3～0.4周的跳变。若去掉这种跳变,其天均值的标准差降为0.03～0.04左右。因此本文建议在求解最终估计UPD的时候,至少需要对经历地影的BLOCK IIA卫星在出地影后额外估计一个UPD参数。(6)发现并指出Carrier-range可改善数据的连续性,进而可提高结果质量。通过比较分析和实例验证,发现相对于传统方法,Carrier-range策略具有更好的数据连续性,其等价于将同一测站同一卫星的所有模糊度均连起来了,从而拥有更好的数据连续性。实验发现采用同样的数据,采用Carrier-range策略得到的轨道的天与天之间的重复轨道RMS值相对传统方法提高约3～4mm。(7)提出了加快无电离层组合观测值数据处理策略,解决了大规模GNSS网数据处理效率低的难题。实验发现：当解算含有460个测站的网时,新策略耗时仅14分钟,而传统策略耗时约82分钟,而且新策略所需的计算时间随着测站数的增长近似呈线性增长,而传统策略耗时随着测站数增长近似呈指数增长。(8)设计了事后和实时高频整数钟差的高效率计算流程,有效提高了事后和实时高频钟差或“整数钟差”计算的效率。实验证明,当采用约250个测站事后计算一天内30s钟差时,耗时仅约32分钟,而当采用约200个测站实时计算钟差时,在所有模糊度固定了的情况下,一个历元耗时不到1s,且直接利用Carrier-range计算的得到的钟差包含了UPD,可直接用于单站模糊度固定。(9)提出了基于原始观测值的数据高效处理方法,有效提高了计算效率,使得采用该方法进行数据日常常规处理成为可能。实验显示,当处理含约100个测站的GNSS网时,在不施加电离层时空约束的情况下,新策略可使单次参数估计时间从97分钟缩短为31分钟,在施加一简单的电离层时空约束后,新策略可使单次参数估计时间从387分钟减少为206分钟。这表明新策略后可有效加速原始观测值的数据处理。
[Abstract]:Global Navigation Satellite System (GNSS) is one of the greatest scientific and technological achievements of the twentieth Century, and has been widely used in many fields such as geodetic, geodynamics and atmospheric research. In recent years, with the full operation of the two generation of the Beidou navigation and positioning system (BDS) in China, Galileo, Europe The continuous development of system (GALILEO) and the modernization of the American GPS system, GNSS has developed from a single system to multiple systems, and dual frequency services develop into a situation of at least three frequency services. This brings great challenges to the current GNSS data processing method. On the one hand, all the public GNSS products before the target are based on the ionospheric combination. The observation value or the data processing method of the difference observation value is obtained. However, this method can not fully meet the requirements of high precision GNSS data processing in the current multi system and multi frequency environment. First, in order to make full use of the advantages of each system, we can get more stable and reliable results, and provide a compatible and unified positive number for other users. In order to keep the information in the original observation value as much as possible, it is necessary to establish the observation equation in the original observation value rather than the difference or the combined observation. On the other hand, the low computing efficiency has been a problem that can not be solved well in the field of GNSS. Even for the GPS system, the method of data processing with combined or differential observation values is difficult or impossible for all the famous GNSS software at present. GNSS networks with hundreds or even thousands of stations are processed for whole network data processing. However, relative to the current combined observation values or differential observation data processing methods, the original observation data processing method is more time-consuming. Even for the GNSS network containing about 100 stations, it is difficult to carry out routine routine processing. Although the first molecular network is solved, The strategy of recombining the result of subnet can alleviate this contradiction to a certain extent, but because the public station of the subnet is used many times, it affects the tightness of the whole network solution. Therefore, this paper proposes a set of strategies to speed up the GNSS data processing, which can not only make the data method based on the ionospheric combination observation value can be processed in a short time. GNSS network with hundreds or even thousands of stations can effectively accelerate the processing of original observation data. The main work and contributions of this paper are as follows: (1) the challenges facing the current GNSS data processing methods are summarized and summarized, and the multi system and multi frequency GNSS data processing model based on the original observation value is systematically studied, and the separation of clock difference parameters is given. The methods of multiple parameters such as system deviation / frequency deviation parameters, DCB parameters and other parameters are discussed. The correlation between ionospheric parameters, DCB parameters and fuzziness is discussed, and the fuzzy degree fixing method under the condition of no ionosphere time and space constraints is discussed. (2) a preliminary discussion on the frequency of GPS and BDS by the processing method of the original observation value is made. The experimental results show that the residual difference of GPS L5 is about 3 times of the L1/L2 residual, and it can reach 9 to 10mm, and its variation and change trend are basically the same in a few days, which means that GPS L5 may contain a certain system deviation from L1/L2 incompatibility. The residual of the three frequencies of BDS is all in 2-3mm and no similar situation is found. The experiment also found the L5 code. The DCB estimation on the pseudo range is very stable in several days or even a month. The standard deviation of the monthly mean value is better than that of 3cm. (3), which evaluates the similarities and differences between the GNSS data processing method based on the original observation value and the product based on the data processing method based on the unionospheric combination values in the condition of using two frequencies without applying the ionosphere constraint. It is found that there is no obvious difference in the orbit, clock difference, troposphere and inter system deviation calculated by the two methods, which indicates that the data method based on the original observation value is equivalent to the data square method without ionosphere combined observation when using two frequencies and does not impose the ionospheric spatiotemporal constraints. (4) the PPP fuzziness technique is proposed. A new method of generating Carrier-range under the unified benchmark is designed and a set of high efficiency GNSS data processing flow is designed on this basis. This paper, starting from the basic observation equation of PPP positioning, proves the feasibility of using the single station ambiguity technology to generate Carrier-range under the unified datum, and designs a complete set of high efficiency GNSS data flow. Six steps should be included: calculating the precision orbit and satellite clock difference of the suitable network with a global and uniform distribution; using the floating-point ambiguity obtained in the orbit determination process to calculate the satellite's UPD, fixed satellite orbit and clock difference, PPP positioning one by one, and using the generated UPD to fix the single station ambiguity, and then the carrier phase view. The measured values are converted into Carrier-range, and they are written into the new RINEX; the two steps are repeated until all stations have generated a new RINEX file; the final net solution is made using a new RINEX file containing Carrier-range. The final network solution no longer needs to be estimated or only needs a small amount of Fuzzy degree, and the calculation efficiency will be greatly improved. Two modes can be adopted, one is only Carrier-range, and the UPD parameter will be absorbed completely by satellite clock difference, no longer need to be estimated, the satellite clock difference is also due to the absorption of UPD, the "integer clock". The other method is to use the pseudo distance and Carrier-range at the same time, in order to make the pseudo distance and Carrier-range clock difference definition consistent, It is necessary to estimate a set of UPD parameters on Carrier-range, at this time the absolute datum of the clock difference is determined by the pseudo distance and is consistent with the IGS definition. (5) the stability of the UPD parameters is analyzed statistically and the UPD parameter estimation method is proposed. By the analysis of the experimental data, it is found that the narrow lane UPD in the BLOCK IIA GPS satellite with non experience shadow is stable in 24h. The standard deviation of the mean balance is mostly not more than 0.05 weeks, and the BLOCK IIA satellite in the experience shadow may have a 0.3 to 0.4 week jump in the narrow lane UPD after the out of earth shadow. If this jump is removed, the standard deviation of the mean value is reduced to about 0.03 to 0.04. Therefore, this paper suggests that at least the BLOCK II of the experience is needed when the final estimate of the UPD is solved. The A satellite estimates an additional UPD parameter after the out of earth film. (6) we find and point out that Carrier-range can improve the continuity of the data and improve the quality of the results. Compared with the traditional methods, it is found that the Carrier-range strategy has better data continuity than the traditional method, which is equivalent to that of the same satellite in the same station. The fuzzy degree is connected, and the data continuity is better. It is found that the same data, the RMS value of the orbit between day and day using the Carrier-range strategy is raised about 3 to 4mm. (7) relative to the traditional method, and the data processing strategy of accelerating the non ionospheric combination observation is accelerated, and the large-scale GNSS network is solved. The experimental results show that the new strategy takes only 14 minutes when solving the network with 460 stations, while the traditional strategy takes about 82 minutes, and the time required for the new strategy increases approximately linearly with the increase of the number of stations, while the traditional strategy consumes an approximate exponential growth with the number of stations. (8) design The efficient calculation process of the post and real time high frequency integer clock difference effectively improves the efficiency of the calculation of the clock difference between the post and the real time high frequency clock or the "integer clock difference". The experiment proves that the time consuming time is only about 32 minutes when about 250 stations are used to calculate the 30s clock difference in one day, and when about 200 stations are used to calculate the clock difference in real time, all the fuzzy degrees are found. Under the fixed condition, a calendar time consuming less than 1s, and the clock difference obtained by Carrier-range calculation directly contains UPD, which can be used directly for the fixed ambiguity of single station. (9) a high efficient data processing method based on the original observation value is proposed, which effectively improves the computational efficiency and makes the routine routine processing of the data by this method. The experiment shows that, when the GNSS network containing about 100 stations is processed, the new strategy can shorten the estimated time of single parameter from 97 minutes to 31 minutes without applying the ionosphere spatiotemporal constraints. After applying a simple ionospheric spatiotemporal constraint, the new strategy can reduce the single parameter estimation time from 387 minutes to 206 minutes. The new strategy can effectively accelerate the data processing of the original observation data.
【学位授予单位】：武汉大学
【学位级别】：博士
【学位授予年份】：2015
【分类号】：P228.4

【相似文献】