基于明考夫斯基距离的冲突证据合成方法  

Conflict evidence DS combination method research based on Minkowski distance

作　　者：张大敏[1] 魏永超[1] 

机构地区：[1]中国民航飞行学院,广汉618307

出　　处：《四川大学学报（自然科学版）》2012年第4期790-794,共5页Journal of Sichuan University(Natural Science Edition)

基　　金：国家自然科学基金(61079022)

摘　　要：提出了一种新的证据预处理的冲突证据合成方法.融合系统中各证据的重要程度可能不同,直接利用公式合成可能得到与实际相悖的结论.利用明考夫斯基距离,根据系数的不同分别得到绝对距离、欧式距离和切比雪夫距离,三个距离表示有一定差异,通过融合三个距离得到证据距离,从而可以较准确表示证据的关系.利用证据距离进一步转化为表示证据权重的权值,利用权值对证据进行预处理,从而可以把冲突证据的合成问题进行转化,再通过改进的合成公式,对预处理后的证据进行合成,得到合成结果.最后通过同其它算法的算例比较,说明新方法的在合成冲突证据的方面的优越性.One novel conflict evidence combination method based on evidence pretreatment was studied. The importance degree of the evidence were different, the directly combination conclusion of them may be opposed to the real. According to the different coefficients of Minkowski distance, absolute distance, Euclidean distance and Chebyshev distance were obtained, which have certain differences. The evidence distance was obtained through integration of the three distances, and represents a more accurate evidence relationship. Evidence distance was further changed into the weight which reflects the importance of evi- dence. According to weight, the basic probabilities were reallocated, and then were used to get the final combination result by the improved combination method. The superiority of the new method is shown by the different algorithms comparison example.

关 键 词：数据融合 证据合成 明考夫斯基 证据距离 权值 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]