一种多源试验条件下惯导误差系数融合的新方法

New method for fusion of inertial navigation system error coefficients with multi-source testing data

为充分利用各种系统级试验信息提高惯导系统误差模型准确性,提出一种多源试验条件下惯导误差系数融合的新方法。根据各试验中误差系数对总误差的贡献比重来确定误差模型中可分离的误差系数。然后将各试验中分离出的误差系数分为独立系数和公共系数分别进行融合。其中,独立系数作为信任值保留,公共系数的加权融合采用以估计方差阵的迹作为代价函数的随机搜索算法来确定其融合权重。利用某型惯导系统车载、机载和火箭橇试验信息进行融合的结果表明,该方法效果较好,可行性强。

A new method for fusion of inertial navigation system（INS） error coefficients is proposed with the aim to efficiently utilize various kinds of the INS testing data.In each test,the error model coefficients are selected according to their weights on navigation errors.Subsequently,the coefficients are separated into two parts,called detached coefficients and common coefficients.Detached coefficients are used as the fusion re-sults directly.The common coefficients are fused with different weights.The weights are found out by ran-dom searching arithmetic with the trace of evaluated variance matrix as cost function.The error coefficients separated from the vehicle-loaded,airborne and rocket sled testing data are used for fusion experiment.The experiment results show that the proposed method could get good performance.