受证据理论启发的传感器数据融合算法  

Data fusion algorithm inspired by evidence theory

作　　者：熊彦铭[1] 杨战平[1] 

机构地区：[1]中国工程物理研究院电子工程研究所,四川绵阳621900

出　　处：《华中科技大学学报（自然科学版）》2011年第10期50-54,共5页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：十一五行业预研项目(426010501;51305060204);中国工程物理研究院重大预研项目(06-0401)

摘　　要：受证据理论的启发,提出了一种新的传感器数据融合算法.借鉴证据理论中的基本信任分配思想和证据组合思想,将所有测量值构成的集合视为辨识框架,并对各个测量值的精度进行判断,在此基础上将各个测量值分别转换为相应的证据;再利用基于冲突分配法的证据组合规则对所有证据进行组合,所得合成证据的Mass函数即为各个测量值的权值分配函数;最后对所有测量值进行加权求和得到融合结果.算例仿真和实际应用均验证了所提方法的有效性.Novel data fusion algorithm inspired by evidence theory was proposed.The concept of basic probability assignment was applied to generate the frame of discernment,and its evidence was generated based on the precision estimate.The combination rule based on conflict assignment method was also applied,and the mass function of the combined evidence was referred as the weight assignment function of all data.Finally the fusion result was obtained by weighted summation.Excellent performance of the proposed algorithm was demonstrated by numerical examples.

关 键 词：证据理论 数据融合 组合规则 基本信任分配 加权系数 收敛速度 

分 类 号：TP212.9[自动化与计算机技术—检测技术与自动化装置]