摘要
从高斯-牛顿迭代的角度对迭代扩展卡尔曼滤波(IEKF)进行分析,提出了一种基于组合牛顿迭代法的改进IEKF算法。该算法通过实时判断每次迭代对状态的逼近程度,采用加权平均的方法确定新的迭代值,继而采用卡尔曼滤波框架对状态进行量测更新。新算法较传统的IEKF具有精度高以及对初值不敏感的优点。实例仿真验证了该算法的有效性。
By analyzing iterated extended Kalman filter（IEKF） in terms of Gauss-Newton iteration,a modified IEKF algorithm was obtained with the globally convergent hybrid Newton′s method.This algorithm was used to determine the new iterated value by means of weighted average as well as real-time judgment of each iteration approaching to the actual state.And then,the Kalman filtering framework was used for the new iterated value to update the state vector.Compared with the conventional IEKF,this algorithm is advantageous to accuracy and robustness.The simulation results prove that the algorithm is effective.
出处
《海军工程大学学报》
CAS
北大核心
2012年第1期15-19,共5页
Journal of Naval University of Engineering
基金
国家自然科学基金资助项目(40904018)
教育部重点实验室开放基金资助项目(201001)
作者简介
常国宾（1984-），男，博士生，主要研究方向为惯性技术及应用、最优估计技术，E-mail：guobinchang1984@gmail．com。
