论文部分内容阅读
针对信息融合中冲突证据组合时易出现的一般冲突、一票否决和鲁棒性等常见问题,有两类改进策略:一类修改DS(Dempster-Shafer)组合规则,另一类修改证据源模型.提出一种基于封闭世界的修改模型方法.引入Jousselme距离函数来量化焦元属性及证据之间的相互关联性,进而计算各证据的支持度.对证据支持度进行加权平均后得到参考证据,利用该参考证据对各原始证据进行不确定性判定,获得各原始证据与参考证据之间的大小相似度和方向相似度.在此基础上建立一个相似度动态修正模型,利用DS组合规则进行证据组合,对动态修正模型的多组组合结果求平均作为最终结果.通过仿真实验验证所提出方法的有效性和合理性.
There are two kinds of improvement strategies for the common problems such as common conflicts, one-vote veto and robustness that are easy to occur when the conflict evidence is combined in the information fusion: one is the Dempster-Shafer combination rule, the other is the modified evidence source model A modified model approach based on the closed world is proposed.Jousselme distance function is introduced to quantify the correlation between focal attributes and evidence, and then the support of each evidence is calculated.With the reference of the evidence support, Use the reference evidence to indeterminate the original evidence to obtain the size similarity and the direction similarity between the original evidence and the reference evidence.On the basis of this, establish a dynamic correction model of similarity and use DS combination rules to make evidence The results show that the proposed method is validated and validated by simulation experiments.