摘要
基于证据理论与直觉模糊集之间的关系,提出了一种新的证据可靠性评估方法,该方法可以在先验知识缺乏的情况下,对各证据源的可靠性进行评估。首先,将证据理论中的基本概率赋值函数(Basic Probability Assignment,BPA)转化为直觉模糊集;然后,通过直觉模糊集之间的相似度度量对各BPA之间的相似度进行计算;在此基础上,提出证据支持度的概念,通过分析证据支持度与证据可靠性之间的关系,获得证据的相对可靠性和绝对可靠性;最后,基于证据折扣运算对原始证据进行修正,采用Dempster组合规则对修正后的证据进行组合。此外,基于直觉模糊框架内的证据可靠性评估,提出了一种多传感器融合方法,通过数值实验对该方法的性能进行了对比分析,结果表明,该方法可以实现对不可靠证据的有效评估。
This paper presented a new evidence reliability evaluation method based on evidence theory and intuitionistic fuzzy sets,which can conduct reliability evaluation for different evidence sources when the prior knowledge is lacked.Firstly,the basic probability assignment(BPA)is transformed to intuitionistic fuzzy sets.Then,the similarity among BPAs is calculated through the similarity measure of intuitionistic fuzzy sets.On this basis,the concept of evidence support degree is proposed,and the relative reliability and absolute reliability of evidence can be obtained by analyzing the relationship between supporting degree and reliability of evidence.Lastly,the original evidence is corrected based on evidence discounting operation,and the corrected evidences are combined by Dempster rule.Besides,this paper proposed a multi-sensor fusion method based on the evidence reliability evaluation considering intuitionistic fuzzy sets.Numerical experiment was conducted to analyze its performance.The results show that this method can effectively evaluate the unreliable evidences.
作者
吴文华
宋亚飞
刘晶
WU Wen-hua;SONG Ya-fei;LIU Jing(Test Training Base,Information and Communication College,National University of Defense Technology,Xi’an 710106,China;Air and Missile Defense College,Air Force Engineering University,Xi’an 710051,China)
出处
《计算机科学》
CSCD
北大核心
2018年第12期160-165,176,共7页
Computer Science
基金
国家自然科学基金项目(61703426
61273275
61573375
61503407
60975026)资助
关键词
证据理论
直觉模糊集
可靠性评估
传感器融合
Evidence theory
Intuitionistic fuzzy sets
Reliability evaluation
Sensor fusion
作者简介
吴文华(1983-),男,硕士,讲师,主要研究方向为指控网络工程与防护、智能信息处理,E-mail:wuwenhua136692@163.com(通信作者);宋亚飞(1988-),男,博士,讲师,主要研究方向为面向态势感知的智能推理与决策;刘晶(1982-),女,硕士,讲师,主要研究方向为指控网络工程与防护。
