舰船组合导航系统的顺序滤波融合算法

Sequential Filtering Fusion Algorithm for Marine Integrated Navigation System

舰船组合导航系统Kalman滤波的集中式融合与分布式融合方法存在滤波精度与计算性能、容错性不可兼顾的缺点。且船舶运动受海洋环境等的影响,其噪声基于白噪声建模过于理想化。针对上述问题,提出了舰船组合导航系统的顺序滤波融合算法。该算法的基本思想是采用一阶Markov过程建立舰船运动的噪声模型,并用状态扩维方法将状态方程转化为符合标准Kalman滤波的基本方程,然后每一时刻检查各导航子系统的有效性,对有效的导航子系统引入顺序滤波融合思想实现导航定位。与传统的顺序滤波融合算法相比,新算法在保留与集中式融合同样高的滤波精度、计算性能好的优点的基础上,新增了实用性强、容错性好的优点。理论分析和舰船SINS/GPS组合导航仿真结果表明了新算法的有效性和优越性。

Neither centralized fusion nor distributed fusion in marine integrated navigation system can have both high filtering precision and good computation performance fault--tolerance. Moreover, it is too idealized to model vessel process noise with white noise neglecting the impacts of marine environments on ship motions. For these problems, a sequential filtering fusion algorithm for marine integrated navigation system is presented. In the new algorithm vessel process noise is modeled with 1st order Markov process, the state equations are converted into the basic equations of standard Kalman filter with state dimension augmentation method, then the effectiveness of each sub navigation system is checked and fusions of all effective data are carried out step by step for positioning the vessel. Compared with traditional sequential filtering fusion algorithms, the new algorithm has the same filtering precision and good computation performance as the centralized fusion algorithm, with the additional advantage of practicability and better fault-tolerance. Theoretical analysis and simulation results of a vessel GPS/SINS integrated navigation system demonstrate the effectiveness and superiority of the new algorithm.