冲突证据融合中证据一致性研究

Research on evidence consistency in conflicting evidence fusion

针对经典Dempster-Shafer证据理论在处理高度冲突证据时易产生合成悖论问题,提出一种基于证据一致性的改进证据理论.该方法利用反映证据一致性的焦元相似度和证据相似度计算证据绝对可信度,并改进原始证据模型.然后,将证据相对可信度和焦元相对识别一致性相结合,建立基于证据一致性的合成规则.仿真结果表明,提出的算法能够合成正常数据和冲突证据,能有效解决0信任悖论、1信任悖论、全冲突悖论和证据失效悖论,与其他同类合成方法相比,合成结果更合理,收敛速度更快.

In order to solve the problem of combination paradoxes generated when classical D-S theory handle the highly conflict evidence,the improved D-S theory based on evidence consistency is proposed.Evidence credibility,computed by similarity of focal elements and similarity of evidence,is taken to revise the original evidence.Then,the evidence combination rules are improved based on evidence consistency,which combines evidence relative credibility and relative recognition consistency of focal elements.Simulation results show that the proposed algorithm can combine normal evidence and conflict evidence,and can solve the problems of 0 probability paradox,1 probability paradox,the whole conflict paradox and evidence failure paradox.Compared with other similar methods,the fusion results are more creditable,and the speed of convergence is faster.