一种基于粗糙集证据理论深度融合的局部冲突快速合成方法  

A Deep Fusion Method Based on Rough Sets and Evidence Theory for Local Conflict Evidence Synthesis

作　　者：倪龙强[1] 张丽华[1] 姚新涛[1] 胡高歌[2] 刘鹏辉[1] NI Longqiang;ZHANG Lihua;YAO Xintao;HU Gaoge;LIU Penghui(Northwest Institute of Mechanical&Engineering,Xianyang 712099,Shaanxi,China;School of Automation,Northwestern Polytechnical University,Xi'an 710072,Shaanxi,China)

机构地区：[1]西北机电工程研究所,陕西咸阳712099 [2]西北工业大学自动化学院,陕西西安710072

出　　处：《兵工学报》2019年第12期2560-2569,共10页Acta Armamentarii

基　　金：军委装备发展部装备预先研究基金项目(61403120205);中央高校基本科研业务费专项项目(3102018zy027)

摘　　要：传统证据理论在进行多焦元属性融合时可能产生组合爆炸,当待合成证据体冲突时容易导致一票否决。针对以上两个问题,通过将待合成证据体转换为知识信息系统,应用粗糙集理论对多源信息进行规则约简,得到约简后的知识规则信息系统,计算出信息系统对每条证据体的支持度;在证据合成中,基于规则信息系统引入证据体支持度、焦元可信度、焦元一致性等因素来表示证据体之间的局部冲突,在此基础上给出冲突证据合成规则。通过测试算例对所提方法进行了测试验证,结果表明:该方法具有较快的收敛速度,计算时间消耗能够得到大幅度降低,在冲突证据合成中具有较强的鲁棒性。A local conflict allocation evidence combination rule based on rough sets theory is proposed to reduce the computational complexity and improve the efficiency in conflict evidence combination. In this proposed method,the information system which is consisted of several evidences is reduced by rough sets,and the confidence of belief function is calculated based on the confidence of evidence and single focal elements individually. At last an example is given to prove the efficiency of this proposed method.

关 键 词：粗糙集 证据理论 数据融合 数据挖掘 冲突证据合成 信息系统 

分 类 号：N945.25[自然科学总论—系统科学]