基于最优分组原则的多传感器分组加权融合算法

多传感器对某一状态进行测量时,将多传感器分组后进行加权融合处理,可实现对待测状态较为准确的估计,但多传感器的组合方式不同,估计效果的差异很大.本文针对分组融合算法中如何选择传感器的分组组合方式进行了研究,依据分组中传感器的个数和各组间传感器测量方差的大小关系,提出了使融合估计值最优的分组原则.并将这些原则加以数学抽象,从理论上证明了基于最优分组原则的分组融合算法的最小方差特性.最优分组原则为评价和选择传感器的分组组合方式提供了一个理论依据.实例分析表明:该分组原则极大地缩小了寻找最优分组组合方式的范围,有效地解决了在如何分组问题上的盲目性,从而使得多传感器测量时的在线优化分组得以实现.

Grouping weighted fusion algorithm for multi-sensor and its optimal grouping principle

When measuring a certain state with multi-sensor, the multi-sensor can be divided into groups and then processed by grouping weighted fusion algorithm, which can obtain relatively accurate estimation of the state to be measured. However the estimation effects are different greatly due to different combination ways of multi-sensor. In this paper, how to divide the sensors properly in grouping fusion algorithm is discussed, and the grouping principle is summarized based on the number of sensors in each group and the magnitude relationship of the measurement noise variance of the sensors in different groups. Using the proposed principle the estimated fusing value can reach an optimal point. Then, we extract the mathematical model of this principle, and prove theoretically the least variance characteristic of the grouping fusion algorithm, which is based on the optimal grouping principle. The optimal grouping principle provides a theoretical basis for evaluating and choosing the ways of grouping properly. Finally, an example is presented, which indicates that optimal grouping principle greatly shortens the areas in searching the optimal ways of grouping and can effectively solve the blindness about the problem of how to group, accordingly optimal online grouping can be realized by this grouping principle when multi-sensor is used in measurement task.