一种改进的多传感器伪序贯滤波算法

An Improved Algorithm of Pseudo-sequential Filtering of Multi-sensor System

传统的多传感器数据融合算法假定传感器之间的测量噪声是不相关的 ,但实际上测量噪声存在着一定的相关性 ,因而会引起滤波精度的损失 .针对该问题 ,文中研究了测量噪声相关情况下的同步多传感器跟踪系统的测量融合技术 .在测量噪声相关的条件下 ,根据线性无偏最小方差估计理论 ,提出了一种改进的同步多传感器伪序贯滤波算法 ,该算法不但适用于噪声不相关情况 ,而且也适用于噪声相关情况 .经仿真研究表明 ,该算法明显提高了航迹的融合精度 :在测量噪声相关时 ,融合精度比传统算法有明显提高 ;而测量噪声不相关时 ,性能与传统的数据融合算法相同 .

The classical fusion algorithm of multi-sensor assumes that the measurement noises across the sensors are not correlated, but it is not true for some practical situation. Aiming at this problem, the measurement fusion technique of synchronized multi-sensor tracking system in case of correlated measurement noises is studied. Under the condition of correlated measurement noises, an improved algorithm of pseudo-sequential filtering of measurement fusion in synchronized multi-sensor system is present based on the linear unbiased minimum variance estimation theory. The algorithm presented in this paper is suitable for the situation of both correlated measurement noise and irrelated measurement noise. Simulation study shows that this algorithm improves the estimation accuracy evidently when the measurement noises are correlated. The results of simulation express its validity.