考虑传感器置信度的改进D-S证据合成算法  

Improved D-S evidence combination rule based on reliability of sensors

作　　者：李昌玺[1,2] 周焰[1] 林菡[3] 郭戈[1] LI Changxi ZHOU Yan LIN Han GUO Ge(Air Force Early Warning Academy, Wuhan 430019, China Unit No.66132 of PLA,Beijing 100043,China Dongfang College, Fujian Agriculture and Forestry University, Fuzhou 350017, China)

机构地区：[1]空军预警学院 [2]解放军第66132部队 [3]福建农林大学东方学院

出　　处：《解放军理工大学学报（自然科学版）》2017年第1期81-86,共6页Journal of PLA University of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目(61401503)

摘　　要：为了解决多源信息融合识别算法中传统Dempster-Shafer(D-S)证据理论极少考虑传感器对融合结果产生影响的问题,提出一种考虑传感器置信度的D-S证据合成算法。首先建立传感器与参考传感器关于任一特征的置信距离,通过置信距离修正传统灰关联分析法的关联系数,从而计算传感器置信度;然后通过证据间的相似性定义任一特征所得证据的全局冲突系数,将全局冲突系数与传感器置信度相结合,定义了一种传感器所得证据可用度的计算方法;最后根据证据可用度对传统D-S证据理论进行修正,提出一种新的证据合成规则。仿真及实验数据表明,相比已有算法,新算法的目标识别率更高,对证据冲突的处理更有效。On target recognition algorithm of multi-source information fusion based on Dempster-Shafer（DS）evidence theory,the influence caused by combination results of sensors is rarely considered.In order to solve this problem,a D-S evidence combination rule considering the reliability of sensors was proposed.First,the confidence distance between the sensor and the reference sensor about any feature was established,the correlation coefficient of the traditional gray relational analysis method by confidence distance modified,and then the reliability of sensors calculated.Secondly,the global conflict coefficient was defined by any obtained evidence by similarity between evidences,and then the reliability of evidences was got by combining the reliability and global conflict coefficient.Finally,the D-S evidence theory was corrected according to the reliability of evidences and a new evidence combination rule proposed.Experimental data show that compared with pre-existing algorithm,the new method can recognize targets better and deal with conflict evidence more effective.

关 键 词：置信度 D-S证据理论 目标识别 合成算法 

分 类 号：TN951[电子电信—信号与信息处理]