基于落影相似度的证据合成规则

在多传感器目标决策系统中,由于敌方或者环境的影响,由D-S合成规则,会得出与直觉相悖的结论。提出了一种改进的证据合成规则。首先,利用条件布尔代数距离定义了证据的落影距离,并推出证据的落影相似度;然后,由证据的落影相似度定义证据的可信度,作为证据参与组合时重要程度的权值;最后,按照Murphy的平均加权证据合成规则对证据进行组合。理论分析和数值算例表明：该方法是有效的,且在冲突信息目标决策上优于一些现有方法。

Drop-falling Similarity for Evidence Combination

As for the multi-sensor target decision-making system, with the affection of the opponent and the environment, the use of D-S combination rules in dealing with high conflict information will always arrive at a counter- intuitive conclusion. In this paper, an improved evidence combination rule is proposed. Firstly, conditional Boolean algebra distance is applied to define drop-falling distance, from which drop-falling similarity is deduced. Then, the credibility of evidence, as the weight of significance in evidence combination, is defined from the drop-falling similarity. Finally, the evidences are combined according to Murphy''s combination rule. Both theoretic analysis and numerical examples confirm the validity of the proposed approach and the superiority to the existing methods in combining conflicting evidences.