边缘CPF算法及在重力梯度辅助导航中应用

Marginal cubature particle filter and its application in gravity gradient aided navigation

惯导固有原因使得载体长时间航行累积大量误差。可通过重力梯度量测与惯导组合导航方法来修正导航误差。先对重力梯度仪与惯导组合导航原理进行阐述,提出重力梯度仪辅助INS(GAINS)的系统框架图,对导航用重力梯度图和重力梯度仪进行分析,设定组合量测方程。然后根据状态空间方程的特点,提出使用边缘Cubature粒子滤波(CPF)进行融合估值。通过理论方法证明其对方差的减小,同时给出算法流程。相同条件下与已有APO-PF算法仿真进行经纬度RMSE结果对比,表明该算法估值精度更高;并用CEP对导航误差研究,得到在性能较低的惯导条件下、在梯度仪12E和102E噪声下4h的CEP数值分别为0.044 n mile和0.072 n mile的结果。最后对状态方程简化,定性分析出其余状态量的估值效果。

INS has large accumulated errors after long journey owing to its inherent reason, which can be decreased by integrating INS with gravity gradient measurement. In this paper, the theory of gravity gradiometer aided INS is described, and its framework chart is presented. Then the gravity map and gradiometer is analyzed, and its measurement equation is set. Based on the feature of state-apace function, a marginal cubature particle filter is proposed to provide estimation in information fusion, and then its variance is proved to be reduced by statistical theory. In addition, the flowchart of algorithm is given, and the simulation is performed. Under the same condition, the position accuracy of the proposed algorithm is higher than that of the existing APO-PF algorithm based on the comparison of their attitude and longitude's RMSE results. The CEP is used for assessing navigation errors, and it is shown that the gradiometer's 4 h CEPs under 1E2 and 10E2 noises are 0.044 n mile and 0.072 n mile, respectively, in low-accuracy inertial navigation system. Finally, the state equation is simplified and used to qualitatively analyze the estimation effect of other state variables. ©, 2014, Editorial Department of Journal of Chinese Inertial Technology. All right reserved.