Abstract: In order to combine highly conflicting evidence efficiently, a new evidence combination rule making use of evi- dence classfication was proposed based on triangular norm and discount factor analyse. First, utilizing average evidence distance and discount factor based on triangular norm, the evidence was classified into three categories: reliability, no conflict and conflict .The discounting factors of the former two categories of evidences were set to one, which keeps the evidence hold of the right hypothesis to a great extant and makes the fusion results focus onto the right hypothesis more strongly. Then the was improved evidence weight was obtained based on evidence distance, and the modified evidence was obtained by verifying the conflicting evidence according to the weighting rule in order to eliminate the conflict. Fi- nally, according to the Dempster's rule, the modified evidence was combined. Numerical examples show the efficiency and rationality of the proposed approach.

Key words: Evidence theory, Conflict, Discount factor, Triangular norm