摘要
D-S(Dempster-Shafer)证据理论是一种有效的不确定性推理方法,但在组合高冲突证据时,D-S证据理论得到的结果却往往有悖常理。为了解决冲突证据的合成问题,考虑到不同的证据在合成过程中的重要程度不同,提出了一种新的基于证据权重的D-S改进算法。该方法首先引入一个度量证据体间相似度的证据距离函数,建立相应的证据距离矩阵,求出系统中各证据到证据集的平均欧式距离,然后通过信任函数来获得描述各证据重要程度的权重系数并对证据源进行修正,最后利用D-S组合规则对修正后的证据进行合成。通过算例的分析以及与其它改进算法的比较,验证了新方法的有效性和优越性。
Dempster-Shafer evidence theory is an effective method for dealing with uncertainty problems, but the results obtained are counterintuitive when the evidences highly conflict with each other. According to the importance of the evidences, a new improved D-S algorithm based on the weight of evidence is presented to solve this problem. Firstly, a distance function between the bodies of evidences is introduced, the average Euclidean distance between evidence and evidence subset is obtained after establishing the evidence distance matrix, and then modify the evidence through the belief function to get weight coefficients which describe the importance degree of evi- dence, finally these modified evidences are combined together according to the D-S rule. Through the numerical study and compared with the other improved methods verify the validity and superiority of the new method.
出处
《计算机技术与发展》
2013年第5期58-62,共5页
Computer Technology and Development
基金
国家自然科学基金资助项目(41140026)
太原市科技局大学生创新创业专题(120164058)
关键词
D—S证据理论
组合规则
证据距离函数
冲突证据
D-S evidence theory
combination rule
evidence distance function
conflict evidence
