【作者】 李变；

【导师】 胡永辉； 高玉平；

【作者基本信息】 中国科学院研究生院（国家授时中心） ， 天体测量与天体力学， 2011， 博士

【摘要】 机动时频系统是地面固定时频系统的备份和补充,在战时或特殊情况下,替代固定时频系统提供高精度的时间频率信号,满足导航系统各类导航业务的需求。机动时频系统除具有与地面固定时频系统相似的功能外,由于机动条件下的工作环境较为恶劣、原子钟数量少,这就对时间保持方法提出了更高的要求。机动条件下,时间保持方法主要要解决以下几个方面的问题:1.综合时间尺度是机动时频系统时间产生的依据,而且机动条件下原子钟数量少,因此,研究适合的时间尺度算法是机动时频系统时间保持中首先要解决的问题;2.由外界温度变化及震动等引起的噪声、频率跳变、相位跳变及频率漂移等问题;3.为了保证在工作环境差和系统规模受限制等诸多不利因素的条件下,仍然能够提供高精度的时间频率信号,需要设计出一套机动条件下的系统时间监控方法。论文针对以上问题,研究了机动条件下的综合时间尺度产生算法、基于HHT的消噪方法和奇异点检测方法、频率预报方法和主钟监控方法,同时研究了时频基准配置等关键技术,利用中国科学院时间频率基准重点实验室的原子钟资源和相关时频设备,建立了时间保持实验系统,并对机动条件下系统时间的产生和保持进行仿真实验。论文的主要内容包括:研究了机动时频系统中综合时间尺度TAs的产生算法。针对机动条件下钟数量少的特点,对ALGOS算法中的权重计算和钟异常检测作了改进,提出了改进的ALGOS算法。为了满足机动条件下,对综合时间尺度TAs实时性和稳定性的要求,提出了改进ALGOS方法与Kalman算法相结合的算法。研究了噪声平滑滤波方法。机动条件下的工作环境差,使原子钟信号噪声变大、出现频率跳变和相位跳变等问题,因此,在建立时间尺度之前,必须对信号进行消噪和奇异点检测等预处理。论文首次将HHT中的时空滤波器,用于机动条件下,恶劣工作环境引起的原子钟噪声的消噪。研究了机动条件下的时间监控方法。主钟频率源频率预报方法是系统时间监控的关键,根据原子钟的统计特性,提出了基于EMD分解的原子钟频率预报新方法,该方法比传统的频率预报方法更高效、更准确。通过分析原子钟的稳定度指标、主备主钟切换方案、数据采集及对环境的要求,根据机动时频系统的功能和特点,提出了机动时频基准的配置方案。最后,建立了机动时间保持仿真实验系统,开发了系统时间自动监控软件,并利用该软件实现了不同时间保持模式下,系统时间ST的自动监控,达到了预期指标。创新点:1.根据机动时频系统的任务与功能,将改进ALGOS方法与Kalman算法相结合,产生综合时间尺度TAs。改进ALGOS方法解决了原子时Kalman算法的发散性问题,同时Kalman算法弥补了改进ALGOS方法τ<30d时稳定度较差的缺陷。2.时空滤波器基于局部特征时间尺度参数,不需要人为指定中心频率、带宽等各种参数,避免了人为因素的影响。首次将时空滤波器用于原子钟噪声的平滑滤波,并根据远程时间比对结果的特征,提出了时空滤波器+Vondrak平滑的消噪方法,提高了比对结果的精度。3.频率预报方法取决于钟的统计特性和有效的预报间隔。当τ≤10d时,噪声主要表现为白色调频噪声,利用EMD可以提取信号趋势项的功能,提出了基于EMD分解的原子钟频率预报新方法。该方法比目前使用的频率预报方法更准确。更多还原

【Abstract】 The mobile time and frequency system is necessary backup and supplement of the ground time and frequency system in the navigation system. It takes on main tasks of providing high precision time and frequency signals replacing the motive ground station in wartime or under special cases so as to meet all kinds of navigation business demands. It has the similar functions to the ground time and frequency system. However, its adverse work environment and small quantities set high demands to time-keeping methods.There are main several aspects need solving as follows: 1. An integrated timescale is the basis to establish the system time for motive application. While the number of clock is small. So study on an integrated timescale algorithm is the first problem need to be settled. 2. The problems of clock’s noise, signal frequency or phase jumped and frequency shift are caused by outside the temperature change, vibration and so on. 3. In order to ensure the time-frequency system providing high precision time and frequency signals for motive ground station in the disadvantage conditions of the worse working environment and limited system size. The monitor means of system time for mobile application need to be researched.For the above questions, this paper researches several core technologies of an integrated timescale algorithm, de-noising and singularity detecting measures based on HHT, frequency predicting and main clock monitoring techniques and configuration of time and frequency standard in motive condition. Based on atomic clocks and related time-frequency equipments at the Key Laboratory of Time and Frequency Primary Standards, Chinese Academy of Sciences, a time keeping simulation system of the experiment is established. And methods for producing and keeping the motive time-frequency system time is fufilled. The major contents include:An integrated timescale algorithm for motive time-frequency system is studied. For feature of clock small size in motive time-frequency system, the measures of weight calculation and clock abnormal detection in ALGOS algorithm are improved. The improve ALGOS algorithm which raisesτ≥5d stability of integrated timescale is put forward. In order to remedy a defect of poor short-term stability, a combinative algorithm between improved ALGOS and Kalman is brought upward.Research on the de-noise methods. The space-time filter in HHT is firstly used to solve clock noise caused by bad working conditions.Research on monitoring measures of the system time in motive time and frequency system. A new clock frequency predicting means based on EMD extracting signal tendency is raised. It is more efficient and more accurate than traditional one. The frequency prediction problem is solved. According to characteristics of mobile time-frequency system, the configurationplan of the time-frequency standard is proposed by analyzing clock performance, main and stand-by master clock switching scheme, data acquisition and demands for environment.The time keeping simulation system of the experiment is established. And the automatic monitor software of system is developed. They are used to do automatic monitoring experiments of the system time in different time keeping models. And the results achieve the desired targets.Innovations:In the light of features of motive time-frequency system time, an integrated timescale TAs is calculated with the combinative algorithm between improved ALGOS and Kalman, in which the former restrains the latter’s diverge while the latter improves the short-term stability of the former.The time-space filter is based on local character timescale parameter, and don’t need specify the centre frequency, bandwidth such parameters. So some artificial factors are avoided. It is firstly used to filter and reduce atomic clock noises. And in terms of the time comparison result’s feature, a de-noise means of time-space filter + Vondrak is presented which raises precision of comparison result.The frequency predicting method depends on atomic clock statistical property and valid forecast interval. Whenτ≤10d, the main clock noise is white FM. The function of EMD extracting trend of the signal is used to predict clock frequency. This measure is more accurate than present use.更多还原