基于证据方差的加权证据组合  

Weighted Combination of Conflicting Evidence Based on Evidence Variance

作　　者：韩德强[1] 韩崇昭[1] 邓勇[2,3] 杨艺[1] 

机构地区：[1]西安交通大学电信学院综合自动化研究所,陕西西安710049 [2]上海交通大学电子信息学院,上海200240 [3]西南大学计算机与信息科学学院,重庆400715

出　　处：《电子学报》2011年第A03期153-157,共5页Acta Electronica Sinica

基　　金：国家973重点基础研究发展规划(No.2007CB311006);国家自然科学基金(No.60874105;No.60904099);教育部新世纪优秀人才支持计划(No.NCET-08-0345);上海市青年科技启明星计划(No.09QA1402900);航空科学基金(No.20090557004;No.20095153022);上海交通大学"晨星学者计划"(No.T241460612);陕西省电子信息系统综合集成重点实验室项目(No.201101Y17)

摘　　要：DS证据理论在决策级信息融合中有着广泛应用.针对证据组合时的某些反直观结果问题提出一种新的基于证据方差的序贯式加权证据组合方法.首先基于Jousselme证据距离定义了证据方差.每一步证据组合时,依据当前既有证据组合结果序列的方差及当前步新证据加入后的序列方差来生成权重.基于所获权重修正新到证据及前一步组合结果,最后利用Dempster规则完成当前步证据组合.算例分析表明所提方法是合理有效的.Dempster-Shafer evidence theory is widely used in the fields of decision-level information fusion.To suppress the counter-intuitive results encountered when using Dempster＇s rule of combination,a modified sequential weighted evidence combination approach based on variance of evidence is proposed.According to Jousselme＇s distance,variance of bodies of evidence is defined.In each combination step,the weights are generated based on the variances of the sequences of available evidence combination results before and after adding new arriving body of evidence.Then the weights generated are used to modify the bodies of evidence including the previous combination result and the new arriving body of evidence at current step.Finally,according the Dempster＇s rule of combination,the weighted average combination result can be obtained.Some numerical examples provided show the efficiency and rationality of the proposed approach.

关 键 词：证据理论 冲突 证据组合 证据距离 证据方差 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]