基于不确定性度量的证据组合方法

Evidence combination approach based on uncertainty measure

针对Dempster组合规则不能有效组合冲突证据,已有的基于证据间距离的改进组合方法计算复杂度较大的情况,提出了一种证据加权平均组合方法。首先以邓勇等人的组合方法为例计算了基于证据间距离的改进组合方法的计算复杂度,分析了造成计算复杂度较大的原因;然后通过引入证据的不确定性度量概念来描述证据的不确定性并以此为基础定义证据的权重;最后给出算法步骤。理论分析和数值算例表明,该方法能有效融合冲突证据,收敛速度快且降低了计算复杂度。

Focusing on the problem that the Dempster＇s combination rule cannot effectively combine conflict evidences and the high computational complexity of the present solution method based on distance between evidences, a weighted average evidence combination approach was proposed. First, the computational complexity of combination rule proposed by Deng Yong, taken as an instance of combination rules based on distance between evidences, was calculated and analyzed. Then the concept of uncertainty measure of evidence was used to describe the uncertainty of evidence body and as a basis of determining the weight of evidence. The numerical example shows that the evidence combination approach based on uncertainty measure can efficiently combine evidences, whether they conflict with each other or not, with faster convergence speed as well as lower computational complexity compared with those based on the distance between the evidences.