一类噪声相关系统的新型序贯融合滤波

A Novel Sequential Fusion Filtering for Systems with Noise Correlations

提出了一种新的低维序贯式融合滤波算法,有效地解决了多传感器系统中观测噪声与过程噪声两步相关及观测噪声之间两步相关的问题.算法主要基于正交变换的思想,首先经过两次等价的改写观测方程,去除噪声之间的相关性,然后用序贯滤波的思想,依次处理到达融合中心的等价观测信息,从而得到一种新型序贯式融合滤波算法.同时还推导了上述噪声相关情况下的测量值扩维融合算法.通过仿真实验,并与测量值扩维融合算法对比,证明了算法的最优性.

This paper proposes a novel low-dimension sequential fusion filtering algorithm for the multi-sensor fusion filtering problem with two system noise correlations：process noise and measurement noises with two steps cross-correlations,the measurement noises with two steps auto-correlations.Based on the orthogonal transformation technology,measurements can be equivalently transformed as new forms with noise uncorrelated.Then,the measurements can be dealt with according to their arriving sequence to get a real time sequential fusion filter method.Moreover,the optimal centralized fusion filter was also deduced.The simulation in the end verifies the optimality.