基于方差分量估计的正则化配置法及其在多源重力数据融合中的应用  

Normalized Collocation Based on Variance Component Estimate and Its Application in Multi-source Gravity Data Fusion

作　　者：孙文 吴晓平[2] 王庆宾[2] 刘晓刚[3,4,5] 朱志大 

机构地区：[1]61243部队 [2]信息工程大学地理空间信息学院 [3]西安测绘研究所 [4]地理信息工程国家重点实验室 [5]武汉大学地球空间环境与大地测量教育部重点实验室 [6]61206部队

出　　处：《武汉大学学报（信息科学版）》2016年第8期1087-1092,共6页Geomatics and Information Science of Wuhan University

基　　金：国家863计划(2013AA122502);国家自然科学基金(41274029;41304022;41404020);地球空间环境和大地测量教育部重点实验室开放基金(11-01-03)~~

摘　　要：针对信号与误差的方差分量不一致问题及协方差阵病态性问题,分别在多源重力数据最小二乘配置融合过程中引入方差分量估计方法及Tikhonov正则化方法,得到基于方差分量估计的正则化配置法,实际算例结果表明,利用该方法能够有效削弱上述问题,减小重力数据融合结果的系统差,提高数据融合的精度及可靠性。Aiming to address the problem of variance component inconformity between signal and noise from multi-source gravity anomalies,the variance component estimate was introduced into the least square collocation（LSC）method,while the Tikhonov normalizing method was used to address the covariance matrix singularity problem.Experiments under different error conditions were carried out and compared.Results showed that,as compared to the LSC method,the proposed method reduced the systematic error of data fusion.It showed better behaviors both in fusion precision and reliability than standard LSC.

关 键 词：最小二乘配置 方差分量估计 TIKHONOV正则化 多源重力数据 融合 

分 类 号：P223[天文地球—大地测量学与测量工程]