一种基于证据理论的数据融合算法  

A data fusion algorithm based on evidence theory

作　　者：王文庆[1] 杨远玲[1] 杨春杰[1] 

机构地区：[1]西安邮电大学自动化学院,西安710121

出　　处：《控制与决策》2013年第9期1427-1430,共4页Control and Decision

基　　金：国家自然科学基金项目(61100165/F020508);陕西省自然科学基金项目(2007F18)

摘　　要：针对传感器测量值存在系统误差的情况,基于证据理论的思想,提出一种新的数据融合算法.该算法首先将所有测量值根据其与真值的偏差进行分组,并分配不同的基本信任;然后将其构成的集合视为辨识框架,进而将各个测量值转换为相应的证据并进行证据组合,所得合成证据的Mass函数即为各个测量值的权值分配函数;最后对所得分组融合测量值进行加权求和,即得融合结果.仿真结果验证了该算法的有效性.According to the systematic error of sensors data,a new data fusion algorithm is proposed based on the evidence theory.The proposed algorithm divides sensor data into groups according to the deviation,and the concept of basic probability assignment is applied to generate the frame of discernment.The combination rule is also applied,and the Mass function of the combined evidence is referred as the weight assignment function of all data.Finally,the fusion result is obtained by weighted summation.Simulation results show the effectiveness of the proposed algorithm.

关 键 词：数据融合 证据理论 基本信任分配 证据组合 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]