新冲突衡量标准下的证据组合方法  

A Method of Evidence Combination Under the New Conflict Measurement Standard

作　　者：吴克凤 肖建于[1] 宋香鹏 汤祁 WU Kefeng;XIAO Jianyu;SONG Xiangpeng;TANG Qi(School of Computer Science and Technology,Huaibei Normal University,Huaibei,Anhui 235000,China)

机构地区：[1]淮北师范大学计算机科学与技术学院,安徽淮北235000

出　　处：《宜宾学院学报》2023年第6期25-31,共7页Journal of Yibin University

基　　金：国家自然科学基金项目(62006091);安徽省自然科学基金项目(1908085QF286)。

摘　　要：为了解决在使用D-S证据理论对高冲突证据进行融合时产生与实际不符的问题,从修改证据源角度出发,将单子集下的Pignistic概率函数与Pearson相关系数结合,提出一种新的证据冲突衡量标准和基于该标准的加权证据组合方法.先用新的证据冲突衡量标准对不同证据之间的冲突程度进行度量,求出证据之间的相关性测度;然后用相关性测度构建相关性矩阵得到证据之间的支持度,并通过权重系数对证据进行修正;最后用Dempster组合规则对证据进行融合.数值算例表明,基于新的冲突衡量标准下的组合方法能够有效地衡量证据冲突程度,且得到的融合结果比其他相关算法具有更好的精确性和收敛性.To solve the problem of producing inconsistency with reality when using D-S evidence theory to fuse highly conflict⁃ing evidence,a new evidence conflict measure and a weighted evidence combination method based on it were proposed by com⁃bining the Pignistic probability function under a single subset with the Pearson correlation coefficient.The new evidence conflict measure was used to measure the degree of conflict between different pieces of evidence and to find the correlation measure be⁃tween the evidence;afterward,the correlation measure was used to construct a correlation matrix to obtain the support between the evidence,and the evidence was corrected by the weight coefficient;finally,the evidence was then fused by Dempster􀆳s combi⁃nation rule.The numerical examples show that the combination method based on the new conflict measure can effectively mea⁃sure the degree of evidence conflict,and the obtained fusion results have better accuracy and convergence than other correlation algorithms.

关 键 词：D-S证据理论 冲突衡量 Pignistic概率函数 相关系数 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]