结合贝叶斯近似变换的冲突证据合成方法

摘要

为了解决D-S证据理论在处理冲突证据合成问题中存在的不足,改进了证据合成规则,提出了用该规则和贝叶斯近似变换相结合的证据合成方法.首先分析了封闭世界和开放世界假定中识别框架上认知集合间的转换关系,然后根据证据理论和贝叶斯近似变换的基本知识分析冲突产生的影响,最后根据原证据焦元交集和并集的基数关系改进了合成规则,并进行了实例验证.改进的合成规则能更好的保留原来的信息,而且通过与贝叶斯近似变换的结合减少了焦元的数量,结果表明该方法是一种准确有效的冲突证据合成方法.%In order to solve the problems emerging in the situation trying to combine conflict evidences using Dempster-Shafer theory,a new combination rule based on Dubois and prade rule is put forward.A method of using the new rule together with Bayesian approximation is proposed to combine evidences from different bodies.Firstly the proposition transformation of discernment frame is discussed both in the open and closed world assumption.Secondly the effect of conflict is shown based on the DempsterShafer theory and Bayesian theorem.And finally the modified rule depending on the cardinal of set from different evidence bodies is proposed and is shown in the numerical example.Conclusion can be drawn from the example that the new combination rule reserves more information from the combined evidences and Bayesian Approximation reduces the number of focused elements,and using both of them is an effective and accurate method in evidence combination.