估计从UUV航行参数的混合坐标系下的EKF算法

在主从式UUV 协作系统中，由于定位和导航的需要，要求尽快估计出从UUV 的航行参数，但通常所用的递推最小二乘（RLS）算法，其初始方位测量对滤波结果影响大且存在收敛速度慢、计算精度低的缺点，难以满足应用需求，而推广卡尔曼滤波（EKF）算法能较好地克服上述问题。在直角坐标系下（CEKF），方位信息与距离信息相互耦合导致初始振荡剧烈，改为混合坐标系（MEKF）后问题得到了极大的改善。最后，通过仿真及现场试验验证了此改进方法的有效性。

In the master-slave UUV (unmanned underwater vehicle) system, the master UUV must estimate the navigation parameters of the slave UUV as soon as possible so as to satisfy the requirements of positioning and navigation. But the commonly used RLS (recursive least square) algorithm is of slow convergence and low accuracy, and the its result is influenced greatly by the initial azimuth, so it is difficult to meet the application needs. The Extended Kalman Filter (EKF) algorithm can solve the above problems. As the position information and distance information are coupled in the rectangular coordinates (EKF in the Cartesian coordinate system), the initial result has a severe concussion. It is greatly improved after using mixed coordinates (EKF in the mixed coordinate system). Finally, the simulation and field experiments show the validity of the improved algorithm.

国家863 计划资助项目（2006AA04Z262）；国家自然科学基金资助项目（60775061）．