一种基于证据理论的数据融合新算法研究  

A New Data Fusion Algorithm Study Based on Evidence Theory

作　　者：魏雪云[1] 廖惜春[1] 

机构地区：[1]五邑大学信息学院,广东江门529020

出　　处：《传感技术学报》2006年第6期2702-2706,共5页Chinese Journal of Sensors and Actuators

基　　金：广东省教育厅自然科学研究项目资助(Z03076)

摘　　要：从决策级层次出发,给出一种新的基于证据理论的多传感数据融合算法.从传感器的重要度和各证据的可信度两方面综合考虑对融合结果的影响,并根据传感网络的实际应用情况给出了反映传感器重要度及证据可信度的融合系数,引入未知项,使新的算法能更好地实现多传感器的数据融合.通过举例与分析表明新算法能较好地解决证据冲突引起的失效问题.This paper proposes an new algorithm for multi-sensor data fusion based on Dempster-Shafer evidence theory, considering the layer of decision. In this algorithm, the important degree of the sensors and the reliability of evidences are concerned to influence the result of fusion. Present a fusion coefficient involving these two facts according to the actual application of the sensors network and use an uncertain part to make new algorithm do the multi-sensor data fusion better. By taking example and analyzing, it proved that the new algorithm can solve the invalidation problem that is brought by conflict evidence well.

关 键 词：数据融合 D-S证据理论 证据冲突 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]