摘要
针对常规的钟差异常值探测方法仅对数据结构进行分析的情况,该文结合卫星钟差特性,通过ICEEMD+卡尔曼滤波(KF)组合对BDS-3钟差一次差分序列进行异常值探测。利用ICEEMD法将其分解后,采用KF法对含有异常值的高频固有模态函数(IMF)分量进行探测。将剔除异常值的高频IMF分量与其余分量叠加即可获取钟差一次差分修正序列。实验结果表明,经过ICEEMD+KF组合处理的BDS-3钟差一次差分序列较绝对中位差(MAD)法与KF法具有更高的平均拟合精度。此外,ICEEMD+KF组合的平均预报精度较MAD法与KF法具有明显提高,在6~24 h的预报区间内分别具有10%和3%左右的提升效果,提升效果随预报区间的延长而逐步降低。
In view of the situation that the conventional clock bias outliers detection method only analyzed the data structure,this paper used ICEEMD+Kalman filtering(KF)combination to detect the outliers of BDS-3 clock bias primary differential sequence.After it was decomposed by ICEEMD method,KF method was used to detect the high-frequency IMF component with outliers.By superimposing the high-frequency IMF component with outliers and other components,the primary differential corrected sequence of clock bias could be obtained.The experimental results showed that the BDS-3 clock bias primary differential sequence processed by ICEEMD+KF had higher average fitting accuracy than median absolute deviation(MAD)method and KF method.In addition,the average prediction accuracy of ICEEMD+KF combination was significantly higher than that of MAD method and KF method.It had an improvement effect of about 10%and 3%respectively in the prediction interval of 6~24 h,and the improvement effect decreased gradually with the extension of the prediction interval.
作者
刘笃学
贺凯飞
丁志鹏
杨金权
徐向
LIU Duxue;HE Kaifei;DING Zhipeng;YANG Jinquan;XU Xiang(College of Oceanography and Space Informatics,China University of Petroleum(East China),Qingdao,Shandong 266580,China)
出处
《测绘科学》
CSCD
北大核心
2022年第10期66-73,共8页
Science of Surveying and Mapping
基金
国家自然科学基金项目(42174021,42074028)
山东省自然科学基金资助项目(ZR2021MD060,ZR2020MD042,ZR2019MD005)
青岛海洋科学与技术国家实验室开放基金项目(QNLM2016ORP0401)
作者简介
刘笃学(1996—),男,山东德州人,硕士研究生,主要研究方向为GNSS海洋定位。E-mail:mimosa310929@163.com;通信作者:贺凯飞,副教授,E-mail:hekaifei983@163.com
