一种抗差的形变数据插补方法  

A interpolation method of deformation monitoring data series

作　　者：刘宁[1,2] 戴吾蛟 刘斌[1,2] 

机构地区：[1]中南大学测绘与遥感科学系,长沙410083 [2]湖南省精密工程测量与形变灾害监测重点实验室,长沙410083

出　　处：《测绘科学》2017年第9期126-131,190,共7页Science of Surveying and Mapping

基　　金：国家"973"项目(2013CB733303);中南大学教师研究基金项目(2014JSJJ003)

摘　　要：针对传统基于空间插值和时间序列上的插值补全形变缺失数据的方法在空间点位分布稀疏、观测值连续缺失以及含有粗差的情况下插补效果不佳的问题,提出了一种基于抗差Kriged Kalman Filter的形变缺失数据插补方法。该方法是一种时空插值的算法,在空间点位分布稀疏时考虑时间上的相关性,在时间上出现连续缺失时考虑其他点位对插补点的影响,以提高插补缺失数据的精度。又将抗差估计融合到Kriged Kalman Filter中以抵抗形变数据中粗差对插补精度的影响。利用模拟数据及天津GPS地面沉降数据进行了实验分析。结果表明:由于该法考虑了监测点的时空相关性以及具有抗差性能,使得插补结果在空间点位稀疏、连续缺失或具有粗差的情况下都具有较高的插补精度。According to the fact that the traditional way of interpolating the deformation missing data based on spatial interpolation and temporal interpolation performs not very well when spatial monitoring points are sparse, data is continuously missed in temporal domain or data is contained by gross error. An new deformation missing data interpolation method based on robust Kriged Kalman filter are pro- posed. This method is a spatio-temporal interpolation; to improve the interpolation accuracy, it considers temporal correlation while spatial points are sparse, and also considers the influence of other points on in- terpolation points while time is continuously missing. It also integrates the robust estimation and Kriged Kalman filter to resist the gross error effect. The interpolation effect is analyzed by using simulated data and TianJin GPS ground subsidence data. The results show that interpolation accuracy is high even while spatial points are sparse, the data is missed at random and continuously or the data is contained by gross error, because the new method has considered the spatio-temporal correlation and has resistance of robust estimation.

关 键 词：缺失数据 插补 Kriged KALMAN FILTER 形变序列 

分 类 号：P208[天文地球—地图制图学与地理信息工程]