Evidence Combination Based on Belief Interval

Abstract

The theory of belief functions is widely used in the field of multi-source information fusion, where the combination rule of evidence is used to combine evidences from different information sources. However, Dempster's rule of combination might have some counter-intuitive behaviors, which hinders its practical applications. In this paper, we propose an indirect evidence combination method using belief intervals of focal elements. The fused belief interval is obtained through some operations for interval, and then the combined evidence is obtained from the fused belief interval with the uncertainty optimization. Results of numerical examples and experiments show that the new evidence combination method is rational and can avoid the counter-intuitive results of Dempster's rule.