基于改进冲突度量的多证据直接融合算法

A Direct Fusion Algorithm for Multiple Pieces of Evidence Based on Improved Conflict Measure

针对Jousselme证据距离函数不能较好描述证据局部冲突和不能对高冲突证据进行准确冲突度量的不足,该文首先提出改进的Jousselme证据距离函数,该函数基于能够较好描述证据之间局部冲突情况的非重合度对Jousselme证据距离函数进行改进,使其冲突度量结果随非重合度取值及其取值范围的变化按适当比例进行变化;其次,基于冲突系数和新改进Jousselme证据距离函数共同构建改进的融合冲突度量函数。在此基础上,对焦元权系数计算式进行改进,并依此对局部多维冲突信息进行按比例分配。理论及应用分析结果表明,新算法是一种适用性广泛且抗干扰性能好的证据融合算法。

In the light of the disadvantages that Jousselme’s evidential distance function can not describe the local conflicting information of evidence well and can not measure the conflict of high conflicting evidence accurately,an improved Jousselme’s evidential distance function is proposed.In the new function,Jousselme’s evidence distance function is improved by using the non-coincidence degree,which can better describe the local conflict of evidence,so that the conflict measure result of evidence varies proportionally with the value of the non-coincidence degree and the scope of its change.Secondly,an improved fusion conflict measure function is constructed based on the conflict coefficient and the new improved Jousselme’s evidential distance function.On this basis,the weight coefficient formula of focal element is improved,and the local multi-dimensional conflicting information is assigned proportionately.Theoretical and application analysis results show that the new algorithm is a kind of evidence fusion algorithm with wide applicability and good anti-jamming performance.