基于关联系数的D_S融合悖论改进算法

Improved combination rule of high conflict evidence based on correlation coefficient

针对Dempster_Shafer证据融合理论在融合高冲突证据源时出现融合悖论的问题,提出了一种通过定义关联系数并构造绝对支持度对原证据进行预处理,然后采用经典的D_S融合方法对证据进行融合的改进方法。通过对关联系数中增长因子取值进行实验和一些经典改进方法进行对比实验,可以看出本文方法不仅能够有效削减证据之间的冲突,也能保持D_S融合理论的优越性,其在融合的准确性和收敛速度上都有很大的提升。

DS combination rule may cause fusion paradox when the evidences are highly conflicted. This paper proposes a new improved combination rule of evidence based on correlation coefficient. This method can not only reduce the conflict within evidence, but also maintain the benefit of traditional DS method. Analysis and comparison for the experiments prove that the method performed better in accuracy and convergence rates than the existing ones.