基于指数函数的证据冲突衡量方法  

Evidence Conflict Measure Method Based on Exponential Function

作　　者：李军伟[1] 周林[1] LI Junwei;ZHOU Lin(School of Computer and Information Engineering, Henan University, Kaifeng 475004, Henan, Chin)

机构地区：[1]河南大学计算机与信息工程学院,河南开封475004

出　　处：《指挥信息系统与技术》2018年第2期43-48,共6页Command Information System and Technology

基　　金：国家自然科学基金(61771006;61703142和U1504611);河南省高等学校重点科研计划(16A510001);2016年度河南省博士后科研;河南省科技发展计划(162102210401);中国科学院光谱成像技术重点实验室(LSIT201711D);河南大学科学研究基金(2015YBZR026)资助项目

摘　　要：证据间冲突程度的有效衡量是解决高冲突证据融合问题的有效途径之一。针对冲突系数无法有效衡量证据间的冲突程度问题,在分析现有冲突衡量方法基础上,提出了一种基于指数函数的证据冲突衡量方法。首先,根据证据间基本概率赋值(BPA)关系将证据分为一致证据和非一致证据2类;然后,根据证据分类结果,综合利用差异性信息和指数函数分别构造冲突衡量系数;最后,用多种冲突衡量方法进行算例分析对比,试验结果表明本文方法可有效衡量证据间冲突程度。The effective measure of the conflict level between the evidences is one of the effective approaches to solve the combination problem of the high conflict evidences. The conflict coeffi cient cannot effectively measure the conflict level between the evidences. Aimed at the above problem, by analyzing the existing methods, an evidence conflict measure method based on the exponential function is proposed. Firstly, according to the basic probability assignment （BPA） relationship between the evidences, the evidence is divided into consistent and inconsistent evi dences. Then, according to the evidence classification result, utilizing the difference information and the exponential function, the conflict measure coefficients are constructed. Finally, the cal culation example is analyzed and contrasted with multiple conflict measure methods. The experi mental result shows that the method can measure the conflict level between the evidences effec tively.

关 键 词：证据理论 冲突衡量 证据冲突 基本概率赋值 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]