多源信息融合中一种新的证据合成算法  

A Novel Combination Rule of Evidence Theory in Multi-Source Information Fusion

作　　者：李昌玺[1] 周焰[1] 王盛超[2] 苏俊生[3] 

机构地区：[1]空军预警学院四系 [2]中国人民解放军94005部队 [3]中国人民解放军92261部队

出　　处：《上海交通大学学报》2016年第7期1125-1131,共7页Journal of Shanghai Jiaotong University

基　　金：国家自然科学基金青年项目(61401503)

摘　　要：针对传统证据理论中,采用冲突系数K衡量证据冲突,只考虑了证据间的非包容性,而未考虑证据间的差异性,提出了一种改进的证据冲突度量方法.该方法通过K和证据间的pignistic概率距离共同衡量证据间的冲突大小,既体现了证据间的非包容性,也反映了证据间的差异性,并给出了该改进方法的信度合成算法过程.通过实例,验证了新算法的有效性与适应性.In the traditional evidence theory, conflict coefficient K was generally used to measure the con- flict of evidence, which only considered the non-inclusive among evidences, ignoring the differences among the evidences. To solve this problem, this paper first proposed a modified measurement method of evidence conflict, which measured the evidence conflict through conflict coefficient K and the pignistic distance be- tween evidences, which not only reflected the non-inclusive among evidences, but also the differences among evidences. Besides, this paper gave the process of combination rule of the proposed method. Typical examples show that the proposed method is effective and adaptable.

关 键 词：证据理论 证据冲突 合成算法 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]