一种新的冲突表征下的异类源信息融合算法  

Heterogeneous sources information fusion algorithm based on new characterization of conflict

作　　者：刘兵[1] 李辉[1] 邢钢[1] 

机构地区：[1]西北工业大学电子信息学院,西安710129

出　　处：《计算机工程与应用》2012年第18期123-126,147,共5页Computer Engineering and Applications

基　　金：国家自然科学基金(No.61171155);西北工业大学研究生创业种子基金(No.Z2012076)

摘　　要：异类源信息融合时常遇到证据高度冲突的情况,此时应用传统Dempster-Shafe(rD-S)证据理论进行融合将出现错误的结果。针对以往冲突证据融合研究中使用冲突系数k表征证据之间的冲突程度所存在的不足,提出了一种新的证据冲突表征方法。在此基础上确定证据间的冲突度和相似度,得到证据的权重。对加权修正后的证据利用D-S证据理论进行融合。算例验证表明该方法可以有效地对冲突证据进行融合,收敛效果较好。Error results may be obtained when the combined evidences provided by heterogeneous sources are in high degree of conflict. Considering the deficiencies of using the conflict coefficient k to represent the conflict degree in previous conflict evidences fusion studies, a new definition of evidence conflict is proposed. The degree of conflict and similarity among evidences are calculated to obtain the evidence weight. The evidences modified by the weight vector are combined using the Dempster-Shafer （D-S） combination rule. Numerical examples show that this method can combine conflict evidences effectively, and have a good convergence.

关 键 词：异类源信息 信息融合 证据理论 证据冲突 冲突表示 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]