基于熵权的多传感器目标识别方法  

Multi-sensor target recognition based on entropy weight

作　　者：万树平[1] 

机构地区：[1]江西财经大学信息管理学院,江西南昌330013

出　　处：《系统工程与电子技术》2009年第3期500-502,510,共4页Systems Engineering and Electronics

基　　金：国家自然科学基金(10626029);江西省自然科学基金(0611082);江西省教育厅科技项目基金(GJJ08350)资助课题

摘　　要：针对具有多个特征指标的多目标识别问题,提出了一种基于熵权的多传感器信息融合方法。该方法根据熵论的基本原理定义熵权,通过引入优异度、次异度的概念,采用相对接近度来评判距离的大小,从而给出目标识别规则。该方法利用熵权避免了特征指标权重选取的主观性,提高了目标识别结果的客观性和准确性。工件识别实例验证了算法的有效性和可操作性。Aimming at the problem of multi-target recognition with many characteristic indexes, a new fusion method for the fuzzy muti sensor data is proposed according to the entropy weight. The method defines the entropy weight according to the theory of entropy, introduces the notations of the optimal and .suboptimal mem bership degrees and uses the relative approach degree to measure the distance. Hence the rule of target recogni- tion is given. The method may avoid the subjectivity of the weight of characteristic indexes and improve the ob jectivity and accuracy of target recognition. The example of parts recognition proves that the method is both ef fective and exercisable.

关 键 词：模糊传感器 数据融合 目标识别 熵权 相对接近度 

分 类 号：TP274.2[自动化与计算机技术—检测技术与自动化装置]