摘要
在多传感器目标识别中,使用经典证据理论处理冲突证据会得到错误的决策。为解决目标识别系统信息冲突的问题,通过对证据理论分析提出了改进的合成公式。新方法充分利用了证据间的冲突,根据证据关于不同目标的冲突程度将证据间的冲突进行重新分配,然后做出识别决策。新合成公式计算简洁,通过目标识别算例对比表明,处理冲突证据时,该方法能得到合理与可靠的融合结果,收敛速度快,能识别正确的目标。
In Multi- sensor Target Identification system,DS evidence theory will generate a wrongconclusion when dealing with the highly conflict evidences. In order to fuse highly conflicting multisensorinformation in the target identification system, an efficient combination rule is proposed based onthe analyses of the DS evidence theory. The new theory make full use of the conflict, and the evidence’sconflicting probability is distributed to every proposition according to its conflict degree with others, thenthe target is identified. The new combination rule is simple. And the numerical example shows thecombination rule will get a reliable and reasonable result when dealing with conflict evidences, and itwas also faster in data convergence. The new combination rule can identify the right target.
作者
王力
白静
Wang Li;Bai Jing(College of Information Engineering,Taiyuan University of Technology,Taiyuan 030024,China)
出处
《科技通报》
北大核心
2016年第7期134-137,183,共5页
Bulletin of Science and Technology
基金
国家自然科学基金(61072087)
山西省科技攻关(社会发展)项目(20120313013-6)
关键词
DS证据理论
证据冲突
信息融合
冲突程度
DS evidence theory
evidences conflict
information fusion
conflict degree
