北斗三频数据预处理及整周模糊度解算研究

发布时间：2018-12-11 17:49

【摘要】：随着全球卫星导航系统的全面发展,中国自主研制开发的中国北斗卫星导航系统(BDS)的发展进程迅速,但面临着其他国家同时发展导航系统的竞争压力,BDS对系统性能和精度的要求越来越高。作为BDS定位核心技术的数据预处理和整周模糊度解算目前亟待提高。围绕这个问题本文做了深入的研究,详细分析了北斗卫星导航系统的定位原理,并针对北斗信号特点介绍了基本观测量及其误差特性。针对目前BDS定位过程中出现的问题进行分析并提出可行的改进方案,包括小周跳的探测及修复问题和整周模糊度解算中出现的观测值矩阵病态问题。针对小周跳难以探测到问题,分析其原因主要是因为频段波长较短且有各种噪声的影响。遂针对这种情况,选择合适的组合系数来构造观测值线性组合来放大波长并消除电离层延迟和观测噪声,并引入EEMD方法对组合观测值进行分解,提取能量值作为特征向量作为BP神经网络的训练样本对网络进行训练,然后用网络去识别组合观测量是否出现周跳,本文根据数据选用了两个系数组合,分别对无周跳和有周跳的组合观测值进行识别,根据周跳探测识别结果选择最优系数组合。最后,用BP神经网络的预测值进行修复周跳。针对整周模糊度解算中由于观测时间短而出现的观测矩阵病态的情况,本文引入了 TIKHONOV正则化原理对观测矩阵进行正则化,减弱观测矩阵的病态性,然后用LAMBDA方法对整周模糊度的浮点解进行搜索,最后固定整周模糊度的值。本文讨论了利用TIKHONOV正则化-LAMBDA方法来解算整周模糊度时的正则化系数的选取问题,利用多组数据的解算结果确定了最优正则化系数的范围。最后,实验验证本文所做的工作对小周跳的探测及修复和整周模糊度解算的有效性,并针对本文工作结果进行了总结与展望。
[Abstract]:With the overall development of the global satellite navigation system, the development of China Beidou Satellite Navigation system (BDS), which is independently developed by China, is developing rapidly, but it faces the competitive pressure from other countries to develop the navigation system at the same time. BDS requires more and more system performance and precision. As the core technology of BDS location, data preprocessing and integer ambiguity resolution need to be improved. This paper makes a deep research on this problem, analyzes the positioning principle of Beidou satellite navigation system in detail, and introduces the basic observations and their error characteristics according to the characteristics of Beidou signal. This paper analyzes the problems in the process of BDS localization and puts forward some feasible improvement schemes, including the detection and repair of small cycle slips and the ill-posed problem of observation matrix in the whole cycle ambiguity solution. In view of the difficulty of detecting the small cycle slip, the main reason is the short wavelength of frequency band and the influence of various noises. In order to solve this problem, a suitable combination coefficient is selected to construct a linear combination of observations to amplify the wavelength and eliminate ionospheric delay and noise, and the EEMD method is introduced to decompose the observed values. The energy value is extracted as the characteristic vector to train the BP neural network, and then the network is used to identify the cycle slip of the combined observations. In this paper, two combinations of coefficients are selected according to the data. The combined observations without cycle slip and with cycle slip are identified, and the optimal combination of coefficients is selected according to the results of cycle slip detection. Finally, the BP neural network is used to repair the cycle slip. In this paper, the TIKHONOV regularization principle is introduced to regularize the observation matrix in order to weaken the ill-condition of the observation matrix, in view of the sick condition of the observation matrix in the whole cycle ambiguity solution because of the short observation time, in this paper, we introduce the principle of TIKHONOV regularization to regularize the observation matrix. Then the floating point solution of integer ambiguity is searched by LAMBDA method, and the value of integer ambiguity is fixed. In this paper, we discuss the selection of regularization coefficients for integer ambiguity calculation by using TIKHONOV regularization-LAMBDA method, and determine the range of optimal regularization coefficients by using the results of the solution of multiple sets of data. Finally, the effectiveness of the work done in this paper to detect and repair the small cycle slip and to solve the integer ambiguity is verified by experiments, and the results of this paper are summarized and prospected.
【学位授予单位】：昆明理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN967.1

【相似文献】