基于多传感器的序贯式融合有限域H∞滤波方法

Sequential Fusion Finite Horizon H∞ Filtering for Multisenor System

与集中式和分布式融合滤波器相比,序贯式融合滤波器不仅保证了估计精度相同,而且在对测量值即到达即滤波、部分测量值缺失等方面都具有灵活性、自适应性和实时性等特点.为此,本文针对一类噪声能量有界的多传感器动态系统,给出了一种序贯式融合有限域H∞滤波器.首先,利用测量值扩维的方法,给出一种集中式融合有限域H∞滤波器;然后,利用H∞滤波的性能指标与二次型不等式之间、以及Hilbert空间二次型的稳定点与Krein空间正交投影之间等的对应关系,构造出一种序贯式融合有限域H∞滤波器;最后,从理论与数值仿真两方面验证了新滤波器与集中式融合有限域H∞滤波器的性能等价性.

Compared with the centralized fusion filter and the distributed one, sequential fusion filter not only works with the same fusion estimation precision, but also with the advantages of flexibility, adaptively, real-time and so on. Therefore, a novel sequential fusion H∞ filter is proposed for the multisensor system with bounded energy noises in this paper. Firstly, utilizing the method of measurement augmented, a centralized fusion finite horizon H∞ filter is given. Then, based on the correspond relationship between the H∞ filtering performance index and the quadratic inequality, and the relationship between the stationary point of the quadratic form in Hilbert space and the projection in Krein space, a novel sequential fusion finite horizon H∞ filter is derived in this paper. Finally, the equivalency of the new filter and the centralized fusion finite horizon H∞ filter is demonstrated from theory analysis and numerical simulation.