An improved evidence classification synthesis method combined information entropy

Abstract

Since the evidence classification synthetic method is not enough accurate in the classification of general evidence and conflict evidence, it leads to the final fusion is not precise. In order to overcome the shortcomings of the D-S evidence classification synthesis rule, this paper redefines evidence classification attributes in the basis of the similarity between evidences and utilizing the employed information entropy attribute. Combining with the similarity attribute, the evidence could be divided into high credibility evidence, general evidence and conflict evidence. According to the division, the evidence set classified is assigned with different importance coefficients, which could be revised to improve. After improving, the general evidence and high conflict evidence are pushed close to the high credibility evidence opinion. Finally, the revised evidence is synthesized with the D-S evidence combination rule. Experimental results show that the improved approach not only solve the conflict issue well, but also reduce the evidence uncertainty. And thus, better data fusion is achieved.