摘要
证据融合是提高目标识别准确性的有效方法.为了解决高度冲突证据融合时产生不合理结果的问题,提出了证据源权重的评定原则.引入了证据距离的概念,根据证据源权重的评定原则,提出了证据源权重的计算方法,实现了对各传感器证据的按权相加修正,从而在信息融合之前消除了冲突证据,避免了不合理结果的产生.为了有效利用先验知识,提高目标识别的效率,分两种情况设计了证据融合的方案,并进行了融合复杂性分析.分别对两种证据融合方案进行了仿真试验并进行了比较分析,仿真结果验证了使用证据融合进行目标识别的有效性.
Evidences fusion is an effective approach for improving target identification.In order to solve the unreasonable results generated by conflictive evidences,the assessment principle for weight coefficients of evidence sources was proposed.The evidence distance concept was introduced and a calculation method for weight coefficients of evidence sources was proposed according to the assessment principle for weight coefficients of evidence sources.Evidences from multiple sensors were modified and added with these weight coefficients.As a result,evidences conflict is dissolved and unreasonable results are avoided.For the purpose of efficiently utilizing the transcendent knowledge and accelerating target identification,two evidence-fusion schemes were designed and the fusion complexation was analyzed.Simulations were carried out to test the two evidence-fusion schemes,and the results verify the validity of the schemes in target identification.
出处
《北京航空航天大学学报》
EI
CAS
CSCD
北大核心
2010年第11期1365-1368,共4页
Journal of Beijing University of Aeronautics and Astronautics
基金
总装备部装备预研重点基金资助项目(9140A04040106HT0801)
关键词
传感器网络
信息融合
复杂性
sensor networks
information fusion
complexation