证据合成的一般框架及高度冲突证据合成方法  

General framework for evidence combination and its approach to highly conflicting evidence fusion

作　　者：卢正才[1] 覃征[1] 

机构地区：[1]清华大学计算机科学与技术系,北京100084

出　　处：《清华大学学报（自然科学版）》2011年第11期1701-1705,1716,共6页Journal of Tsinghua University(Science and Technology)

基　　金：国防"十一五"预研基金项目(402040202)

摘　　要：针对D-S证据理论在合成高度冲突证据时会得到有悖常理的无效结果的问题,提出了有效解决冲突须同时修正证据源和合成规则的原则,并给出了证据合成的一般框架;在此框架下,提出了一种新的证据合成方法。该方法首先对证据源进行两级可靠性度量处理,然后再按照合成规则进行合成,最后采用基于Pignistic概率距离计算的权重把冲突系数分配给各焦元,实现对合成结果的合理修正。实验结果表明:该方法能有效地解决高度冲突证据的合成问题,与其他合成方法相比,该方法合成效果更好。Counterintuitive and invalid conclusions are produced by the D-S theory of evidence with highly conflicting evidence. This paper gives a principle for effective problem solving when both evidence sources and combination rules have to be improved together, with a general framework for evidence fusion. First, two-level reliability measurements are applied to deal with evidence sources. After that, the improved evidence sources are fused by the combination rules. Finally, the conflicting coefficient is apportioned to focal elements based on the weights calculated from the Pignistie probability distance, which gives reasonable corrections for the fused results. Tests demonstrate that the approach more effectively settles the problem caused by highly conflicting evidence fusion than other methods.

关 键 词：信息融合 证据理论 证据冲突 合成规则 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]