Abstract:

The evidence theory is widely applied in artificial intelligence and information fusion fields as a kind of indefinite inference, however, the Dempster combination rule is not reasonable when dealing with conflict evidences and its result is also invalid. To solve this problem, we present a combination method of conflict evidences based on each evidence value. Firstly, the standard scale of the evidence value is defined. Secondly, each evidence value is calculated by the Euclidean distance and they are regularized as their evidence weight. Then an uncertain part is introduced to modify the evidences. Finally, the evidences are combined by using the Dempster-Shafer combination method. The analysis of classical examples shows that the proposal is effective and feasible when evidences conflict.

Key words: evidence theory, conflict evidence, evidence value, evidence source modification