摘要
当海况不佳时,水下航行器大幅晃动,捷联惯导系统无法快速完成自主初始对准,因此提出了利用多普勒计程仪提供的速度信息进行运动中辅助对准。针对在非线性对准中扩展卡尔曼滤波存在精度低,且需要计算雅可比矩阵等不足,提出了一种基于非线性预测滤波的求容积卡尔曼滤波算法。该滤波算法将惯性器件测量误差作为模型误差使用非线性预测滤波器进行实时预测,然后再利用求容积卡尔曼滤波对模型误差补偿后的系统进行状态估计。仿真结果表明,与扩展卡尔曼滤波和求容积卡尔曼滤波算法相比,该滤波算法能够不仅提高失准角特别是方位失准角的估计精度,其精度约为45″,而且加快了收敛速度。同时由于该滤波算法降低了系统状态的维数,因此也大大减少了计算量。
The initial alignment of SINS cannot be achieved quickly under terrible ocean environment.In this paper,the velocity information of DVL is used to help implement alignment in motion.In view that the extended Kalman filter in nonlinear alignment has low accuracy and needs to calculate the Jacobian matrix,a method of combining nonlinear predictive filter with cubature Kalman filter is put forward,namely NPF-CKF.The NPF-CKF algorithm takes the measurement error of the inertial measurement unit as the model error,which is estimated online based on NPF,and then the system state is estimated by CKF based on the compensated model.Simulation results shows that,compared with the extended Kalman filter and the cubature Kalman filter,the NPF-CKF filter can not only improve the estimation accuracy of all misalignment angles especially the azimuth one whose precision is within 45,but also make the convergence faster.The proposed algorithm decreases the state dimension,so the computation burden is reduced.
出处
《中国惯性技术学报》
EI
CSCD
北大核心
2011年第6期654-658,共5页
Journal of Chinese Inertial Technology
基金
国家自然科学基金(160040300008)
关键词
动基座
初始对准
非线性预测滤波
求容积卡尔曼滤波
dynamic base
initial alignment
nonlinear predictive filter
cubature Kalman filter
