改进的证据组合方法

Improved Approach for Evidence Combination

针对在空中目标识别中,证据理论不能正确处理高度冲突的证据以及冲突因子度量冲突的方法存在不合理性等问题,提出了改进的证据组合方法。该方法包括冲突度量方法及证据组合规则,基于Jousselme距离的冲突度量方法通过计算证据与证据均值的距离,求出各证据间的冲突;基于Lefevre方法框架的证据组合规则定义了基本概率指派(BPA)中非冲突部分所占的比例,再将冲突部分重新分配。仿真实验与比较结果表明,提出的改进方法较有效地解决了证据理论存在的问题,提高了目标识别结果的可靠性。

Considering the difficulty of conflict evidences combination and the measuring conflict irrationality by evidence theory in target recognition,an improved approach for evidence combination was proposed.The method included the conflict measure and evidence combination rule.The conflict measure based on Jousselme distance got the conflict by calculating the distance between the evidences and the mean of evidences.The evidence combination rule based on the framework of Lefevre approach defined the proportion of the non-conflict in basic probability assignment（BPA）,and the proportion of the conflict is reassigned.By simulations,the proposed approach was compared with the existing approaches.The results show that the proposed approach could solve the problems in evidence theory efficiently and improved the reliability of target recognition results.