GPS星载钟特性分析及短期预报算法研究

发布时间：2018-04-27 01:21

本文选题：GPS + 原子钟　；参考：《长安大学》2017年硕士论文

【摘要】：在全球导航系统中,想要获得精确的定位,那首先要实现精确的时间测定。也就是说,要建立高准确度和高稳定度的原子频标和时间系统成为了高精度导航定位的关键之处。原子钟作为目前世界上最准确的计时工具,同时也是卫星导航系统有效载荷的核心部分,它的稳定性性能直接决定了导航定位和时频传递的精度的提高[8];本文首先对GPS钟差序列进行了长期的特性分析,给出卫星钟的相位、频率、频漂以及稳定度的计算模型。除此之外,由于钟差预报是实现时间同步的重要基础,所以在研究中及实际中,很有必要建立高精度的卫星钟差预报模型,这从另一方面决定了卫星导航定位的准确度及精确度。由此,本文根据原子钟时频特性的相关理论和研究成果,详细探究了星载原子钟时频特性,以及在卫星钟差预报模型中发现的若干值得探索的问题[23]。本文主要研究内容和成果包括:(1)首先对于GPS在轨卫星各类原子钟,有BLOCK IIR-M和BLOCK IIR,BLOCK IIA、BLOCK IIF,采用常用的二次多项式模型拟合得到卫星钟差模型,从而根据相关公式得到了GPS在轨卫星钟的相位、频率、频漂以及残差序列,下一步依次计算了Cs钟和Rb钟的稳定度指标,本文采用的是重叠哈达玛方差,进一步揭示了GPS在轨卫星钟的相位、频率、频漂及残差指标的变化规律;发现GPS BLOCK IIF型卫星钟稳定度最高,其次是GPS BLOCK IIR-M和BLOCK IIR型Rb钟[27],然后是GPS BLOCK IIA型Cs钟。稳定度最差的是GPS BLOCK IIA型Rb钟。(2)在预报6小时的短期预报中,首先使用较少的已知数据,本文采用了24个历元建模,在预报精度方面,二次多项式模型预报效果比线性模型的预报精度和灰色系统模型的预报精度都要差一些;之后使用144个历元作为已知数据建模,不难发现这三种预报模型的预报精度差别不大,任选其中一种方法均可[24]。从另一个角度,通过对比预报24小时的预报精度,可知预报误差随着预报时间的加长而增大。综上所述,灰色系统模型作为预报模型,它的很重要的一个优点就是使用较少的已知数据,而可以得到较高的预报精度。随着预报时间的增长或者预报历元数的增加,灰色模型预报精度明显比二次多项式方法的预报精度好得多,说明二次多项式模型有一个明显的缺点就是它的的误差积累特性[13]。这些初步的、经验型的结论对实时精密单点定位卫星钟差预报具有一定的参考价值和借鉴意义[14]。
[Abstract]:In a global navigation system, accurate time measurement is the first thing to achieve. That is to say, the establishment of high accuracy and high stability atomic frequency standard and time system has become the key point of high precision navigation and positioning. Atomic clock is the most accurate timing tool in the world, and it is also the core part of the payload of satellite navigation system. Its stability performance directly determines the accuracy of navigation positioning and time-frequency transmission [8]. In this paper, the long-term characteristics of the GPS clock difference sequence are analyzed, and the calculation models of the phase, frequency, frequency drift and stability of the satellite clock are given. In addition, because the clock difference prediction is an important basis for time synchronization, it is necessary to establish a high-precision satellite clock difference prediction model in the research and practice. This, on the other hand, determines the accuracy and accuracy of satellite navigation and positioning. Therefore, based on the relevant theories and research results of the time-frequency characteristics of atomic clocks, the time-frequency characteristics of space-borne atomic clocks are discussed in detail, as well as some problems worth exploring in the satellite clock difference prediction model [23]. The main contents and achievements of this paper include: (1) first of all, for various atomic clocks of GPS satellites in orbit, there are BLOCK IIR-M and BLOCK IIR BLOCK IIAA BLOCK IIFs. The satellite clock difference model is obtained by fitting the commonly used quadratic polynomial model. The phase, frequency, frequency drift and residual error sequence of GPS satellite clock in orbit are obtained according to the correlation formula. The stability indexes of Cs clock and RB clock are calculated in turn. The overlapping Hadamard variance is used in this paper. The changes of phase, frequency, frequency drift and residual error of GPS satellite clock in orbit are further revealed. It is found that the stability of GPS BLOCK IIF satellite clock is the highest, followed by GPS BLOCK IIR-M and BLOCK IIR RB clock [27], and then GPS BLOCK IIA Cs clock. The worst stability is the GPS BLOCK IIA RB clock. 2) in the short term prediction of 6 hours, less known data are first used. In this paper, 24 epoch models are used, and the prediction accuracy is obtained. The prediction accuracy of quadratic polynomial model is worse than that of linear model and grey system model. Either method is optional [24]. From another point of view, by comparing the prediction accuracy of 24 hours, we can see that the prediction error increases with the increase of forecast time. In conclusion, as a prediction model, grey system model has the advantage of using less known data and obtaining higher prediction accuracy. With the increase of forecasting time or the number of epochs, the prediction accuracy of grey model is much better than that of quadratic polynomial method. It shows that the quadratic polynomial model has an obvious disadvantage, which is its error accumulation property [13]. These preliminary and empirical conclusions have certain reference value and significance for the prediction of satellite clock difference in real time precise single point positioning [14].
【学位授予单位】：长安大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P228.4

【参考文献】