一种基于证据距离的D-S改进算法

针对高冲突下Dempster证据组合规则失效的问题,定义了描述证据可靠程度的可信度的概念,提出了一种改进的D-S组合算法。该方法首先通过证据距离,建立相似性矩阵,确定证据可信度,对证据进行预处理,再采用Dempster规则进行组合。通过仿真算例并与其它改进方法进行对比,验证了新算法在高冲突证据的情况下,可以有效地减小冲突对组合结果的影响,同时提高了证据组合的收敛速度,降低了决策风险。

An Improved D-S Algorithm Based on Distance of Evidence

To solve the invalidation problem of Dempster-Shafer evidence combination rule with high conflict,the concept of credibility is defined to prescribe reliability of evidence and novel improved D-S algorithm is proposed.Through the distance of evidence,the proposed algorithm first constructs a similarity matrix and obtains credibility of evidence,and then preprocesses evidence through introducing credibility,finally combines preprocessed evidence using Dempster's rule.The numerical study shows that,compared with other methods,the new algorithm can deal with high evidence conflict and reduce influence of conflict evidence effectively,and can accelerate convergence and reduce the decision risk.