基于FS-DS的信息融合方法及其应用  

Information Fusion Method Based on FS-DS and Its Application

作　　者：张秋[1] 孙顺远[1] 梁小凡[1] 徐保国[1] 

机构地区：[1]江南大学物联网工程学院轻工过程先进控制教育部重点实验室,江苏无锡214122

出　　处：《计算机工程》2013年第5期48-52,共5页Computer Engineering

基　　金：国家"863"计划基金资助项目(2007AA10Z241;2007AA100408;2006AA10A301);国家部委基金资助项目;高等学校博士研究生基金资助项目(JUDCF11003)

摘　　要：针对多传感器信息采集系统中的数据不确定性问题,提出一种基于证据理论和模糊集合的多传感器数据融合方法。该方法利用相关性函数定义不确定信息的模糊支持概率,由隶属函数得到各个传感器所测信息的可信度,将支持度和可信度转化为基本概率分配函数,通过D-S证据合成辨别出测量精度较高的传感器。实际应用结果表明,该方法可改善证据理论应用中基本概率分配函数难以确定与多传感器之间相互支持程度计算绝对化的问题,与传统的D-S算法相比,融合结果具有更高的精度和可信度。Aiming at the problem of the data uncertainty in information gathering system, a multi-sensor information fusion method based on fuzzy set and evidence theory is proposed, in which the fuzzy support probability of the uncertain information is defined by the correlation function. Credibility of the information measured by each sensor is obtained by using the membership function. The support and credibility are transformed into basic probability assignment function. The sensors with higher measurement precision are identified by D-S evidence combination. Practical application results show that this method can improve the problems that basic probability assignment function is difficult to be determined and the calculation of degree of mutual support is absolute, and the fusion result has higher accuracy and reliability.

关 键 词：多传感器 数据融合 证据理论 模糊集合 隶属函数 置信距离测度 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]