基于相异度矩阵的冲突证据合成算法  

Conflict evidence combination algorithm based on dissimilarity matrix

作　　者：司成[1,2] 张红旗[1,2] 汪永伟[1,2] 常德显[1,2] 

机构地区：[1]信息工程大学,河南郑州450001 [2]河南省信息安全重点实验室,河南郑州450001

出　　处：《计算机工程与设计》2015年第2期482-486,513,共6页Computer Engineering and Design

基　　金：国家973重点基础研究发展计划基金项目(2011CB311801);国家863高技术研究发展计划基金项目(2012AA012704);郑州市科技领军人才基金项目(131PLJRC644)

摘　　要：为解决经典证据理论在证据冲突程度高的环境下合成容易导致Zadeh悖论的问题,提出一种解决冲突证据合成问题的算法。通过计算证据间的欧几里德距离构造相异度矩阵,计算证据的相异支持度、可信度和修正率,对证据进行修正后利用合成算法进行合成。实验结果表明,该算法可以很好地解决Zadeh悖论,合成结果优于现有一些典型方法。Because the combination of traditional evidence theory causes Zadeh paradox in the environment of high conflict evidence,a combination algorithm for conflicting evidence was proposed.A dissimilarity matrix was built by calculating the Euclidean distance between evidences and the degree of dissimilar support,credibility and correction rate were calculated.The combination algorithm was utilized to combine the corrected evidence.Experiments verify that this algorithm can effectively solve the problem of Zadeh paradox and the combination results are better than that of some existing typical methods.

关 键 词：证据理论 合成算法 冲突 相异度矩阵 信息融合 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术]