期刊名称：International Journal of Hybrid Information Technology
印刷版ISSN：1738-9968
出版年度：2015
卷号：8
期号：12
页码：299-306
DOI：10.14257/ijhit.2015.8.12.22
出版社：SERSC
摘要：Evidence theory is an effective method for uncertainty reasoning, which is widely used in areas like expert system, artificial intelligence, pattern recognition and system decision. But traditional DS combination rule will produce the result contrary to intuition on the condition of high confliction. To solve the problem, this paper proposes a modified method based on the option of distance function and correction of support degree. Firstly, it introduces the Minkowsky distance as distance function of evidences and finds the support degree of each evidence in system, then corrects the support degree on the basic of its distribution, finally, it gets the weighted average of evidence by the normalized evidence credibility, and uses the DS combination rule to synthesis evidences. The simulation results demonstrate the effectiveness and reliability of modified method.
