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

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.