摘要
Dempster-Shafer理论合成高冲突证据时结果不合常理,许多改进方法能较好地解决这一问题,但证据比较一致时,这些方法性能相对于Dempster-Shafer合成规则较差.鉴于此,提出了一种基于证据可信度的证据合成方法.首先建立一个证据距离矩阵,之后求出系统中各证据到证据集的均方欧式距离,通过引入可信度函数来获得证据源的可信度因子ε并对证据源进行修正,最后使用Dempster-Shafer合成公式对修正后的证据加以合成.数值实验结果表明,改进后的方法不仅适用于证据高冲突情况,而且也适用于证据比较一致的情况.
Dempster-Shafer combination results may be not in agreement with the real condition when the evidences are highly conflict; and this problem has been widely investigated by some improved methods. However, those methods are not as good as Dempster-Shafer combination rule when the evidences are consistent. Therefore, a novel evidence combination method is proposed in this paper based on evidence credibility. First, the mean square Euclidian distance between evidence and evidence subset is obtained after establishing the evidence distance matrix, and the credible factor e is calculated to modify the evidence resource correspondingly by a confidence function, and then these modified evidences are combined together according to the D-S rule. Simulation experiments demonstrate the effectiveness of this new improved algorithm when dealing with conflicting evidences and consistent ones.
出处
《控制理论与应用》
EI
CAS
CSCD
北大核心
2009年第7期812-814,共3页
Control Theory & Applications
基金
国家自然科学基金资助项目(50605021)
关键词
证据理论
可信度
冲突证据
一致证据
合成方法
evidence theory
credibility
conflicting evidences
consistent evidences
combination methods
