摘要
针对高度冲突证据的合成问题,分析了传统Dempster组合规则及其改进方法的不足,提出了一种新的冲突证据合成方法。该方法首先计算证据之间的距离矩阵和相似度矩阵,然后计算各证据获得其它证据的支持度,将支持度归一化处理得到各证据的可信度,最后基于可信度采用加权平均的方法将证据冲突概率分配给各个命题。数值算例结果表明,该方法减小了冲突证据在组合规则中的作用,有效地降低了干扰信息的影响,能够得到合理的融合结果。
Based on the analysis of the traditional Dempster combination rule and the existing modified algorithms,an improved combination rule for conflict evidences is proposed.First the distance matrix and the similar matrix are calculated,then the support degree of the evidence from the other one is derived,after that the reliability degree can be obtained by the normalization processing,finally the conflict probability is assigned to every hypothesis according to weighted average of reliability degree.Numerical instances show that the proposed method can effectively decrease the impact of conflict evidence,reduce the influence of interference information and gain a reasonable fusion result.
出处
《系统工程》
CSSCI
CSCD
北大核心
2014年第5期150-153,共4页
Systems Engineering
基金
教育部新世纪优秀人才支持计划项目(NCET-09-0074)
关键词
证据理论
冲突证据
组合规则
信息融合
Evidence Theory
Conflict Evidence
Combination Rule
Information Fusion
