基于证据冲突度的多传感器冲突信息组合方法

摘要

Considering Dempster combination rule's disadvantage in efficiently combining the conflict evidence, this paper presented the improved definition of evidence conflict and computation equation of evidence conflict degree based on conflict coefficient and Jousselme distance and established a new evidence combination method based on evidence conflict degree.This method firstly constructed an evidence support degree matrix.Then computed the evidence weight vector.Finally, used Dempster combination rule to combine the evidence modified by weight vector.A numerical example illustrates the efficiency of the proposed method in combining the conflict evidence.Compared with Dempster combination rule and several typical weighting methods, the proposed method is more efficient.%针对Dempster组合规则在多传感器冲突信息融合方面的不足,提出了改进的证据冲突的定义及基于冲突系数和Jousselme距离的证据冲突度的计算式,给出了一种新的基于证据冲突度的证据加权融合方法.该方法首先利用证据冲突度构造证据相互支持度矩阵,进而计算证据的权重,最后利用Dempster规则对加权修正后的证据进行融合.数值实例表明:该方法可以有效融合高冲突信息,与Dempster组合规则和几种典型的加权证据融合方法相比,具有更快的收敛速度,而且收敛效果更好.