基于证据理论的多传感器加权融合改进方法  

Improved Multi-Sensor Weighted Fusion Method Based on Evidence Theory

作　　者：李军[1] 锁斌[1] 李顺[1] 

机构地区：[1]中国工程物理研究院电子工程研究所

出　　处：《计算机测量与控制》2011年第10期2592-2595,共4页Computer Measurement &Control

基　　金：十一五跨行业重点预研项目(426010501);十一五行业重点预研项目(51305060204)

摘　　要：针对多传感器系统中高冲突证据合成问题,考虑到传感器同时具有固有可靠性与实时可靠性的特点,基于证据理论提出了一种新的多传感器加权融合方法;该方法通过计算证据间的相似矩阵获取证据的后验权重,并结合根据传感器固有可靠性预先分配的先验权重,得到证据体的复合权重,然后据此对原始证据进行加权平均,最后利用D-S证据组合规则合成加权平均后的证据;实例仿真表明,与D-S、Yager、Murphy等方法相比,该方法能够更好地处理高冲突证据,且收敛速度更快。In multi--sensor information fusion system, considering the inherent reliability and random uncertainty of sensors, an im- proved weighted evidence combination method is proposed. Similar matrices between evidences are calculated as posterior weights. Combined it with prior weights, which are assigned on the basis of inherent reliability of sensors, hybrid weights are obtained. Evidences are averaged with these complex weights, and then combined with D-S s rule. Simulation results show that, compare with the method only considering one type of above weights, the proposed method is more effective in the integration of sharp conflict evidences, and more efficient in convergence speed.

关 键 词：证据理论 多传感器融合 复合权重 

分 类 号：TP24[自动化与计算机技术—检测技术与自动化装置]