引用本文: | 龚航,朱祥维,刘增军,等.Huber加权的频率稳定度非相关差分抗差估计方法[J].国防科技大学学报,2017,39(5):26-31.[点击复制] |
GONG Hang,ZHU Xiangwei,LIU Zengjun,et al.Uncorrelated differential robust estimation of frequency stability based on Huber weight function[J].Journal of National University of Defense Technology,2017,39(5):26-31[点击复制] |
|
|
|
本文已被:浏览 9139次 下载 5910次 |
Huber加权的频率稳定度非相关差分抗差估计方法 |
龚航, 朱祥维, 刘增军, 李井源, 孙广富 |
(国防科技大学 电子科学学院, 湖南 长沙 410073)
|
摘要: |
计算原子钟频率稳定度时,钟差观测异常会导致Allan方差出现较大的估计偏差。建立Allan方差的差分估计模型,根据钟差的差分序列统计特性分析了相关差分估计的噪声识别与白化等计算复杂度问题;针对上述难点提出一种基于Huber权函数的非相关差分抗差估计方法,建立一种非相关差分序列的构造方法,有效避免了复杂的噪声识别及白化计算;给出Allan方差的非相关差分抗差估计的推导,并对抗差过程引入的误差累积给出了一种抵消方法;给出完整的抗差估计方案,并利用实测数据进行了实验验证。实验结果表明本方法对相位单点跳变、相位阶跃跳变具有显著的抗差能力,抗差估计可使异常引入的相对偏差由近200%降至10%以内。 |
关键词: 频率稳定度 Allan方差 抗差估计 非相关差分 Huber权函数 |
DOI:10.11887/j.cn.201705005 |
投稿日期:2017-02-24 |
基金项目:国防科技项目基金资助项目(3101121);国家自然科学基金资助项目(61403413) |
|
Uncorrelated differential robust estimation of frequency stability based on Huber weight function |
GONG Hang, ZHU Xiangwei, LIU Zengjun, LI Jingyuan, SUN Guangfu |
(College of Electronic Science, National University of Defense Technology, Changsha 410073, China)
|
Abstract: |
The atomic clock′s frequency stability estimated by Allan variance will significantly deviate from its true value if anomalies occur. An uncorrelated differential robust estimation method based on Huber weight function was proposed. Firstly, a difference estimation model of Allan variance was established, and the computation complexity of noise identification and whitening introduced by correlated difference was analyzed on the basis of the statistical property of the clock difference sequence. In order to solve this problem, a construction model of uncorrelated difference sequence was established to avoid complex noise identification and whitening computation effectively. Based on this, the uncorrelated difference robust estimation method of Allan variance was derived, followed by a counteracted technique of error accumulation introduced by robust estimation. Finally, the integrated robust estimation solution was given, and its effectiveness was verified by experiment. The results show that the method has significance robustness for phase outliers and phase jumps, and the relative deviation introduced by anomalies can be reduced from almost 200% to less than 10%. |
Keywords: frequency stability Allan variance robust estimation uncorrelated difference Huber weight function |
|
|
|
|
|