Application of Correntropy-Based CDKF in Transfer Alignment for Accuracy Enhancement of Airborne Distributed POS

For distributed Position and Orientation System (POS), multi-node motion information are measured mainly by transfer alignment technology. The KF-based non-linear filtering algorithm provides optimal integrated solution for transfer alignment. However, on account of uncertain external environment and flexible lever arm compensation error, the noises of transfer alignment exhibit non-Gaussian property and further cause lower accuracy of distributed POS. In this paper, Correntropy-based CDKF is proposed by combining central difference Kalman filter (CDKF) and maximum correntropy criterion (MCC), which aims to decrease the effect of non-Gaussian noises. We rebuild a linear regression model and define a cost function under MCC to modify measurement information, and finally accurate state estimation is achieved. Meanwhile, the error model of nonlinear transfer alignment is formulated where flexible deformation and lever arm effect are considered, and the attitude and velocity information of main POS are transferred to slave IMU to improve its performance. Based on a semi-physical simulation, the performance of proposed method is proved to outperform the standard CDKF in term of estimation accuracy.