A New Rule and Its Optimization of Combining Evidences

Abstract

Due to the defects of the Dempster-Shafer evidence theory, using Dempster's rule to combine evidences has limitations. Aiming at the problem of one-vote veto in Dempster's rule, and the defects of its improved methods in dealing with conflicting evidences, this paper proposes a new combination rule of evidence. When there is a heavy conflict among the to-be-combined evidence groups, the trustworthiness of the evidence is obtained based on the distance between the evidences, the evidences are corrected, and then conflict evidences are distributed in the focal elements using a new combination rule. When the evidences are almost consistent with each other, the Dempster's rule is chosen to combine evidences. The numerical examples show that the new combination rule of evidence makes the best use of the concordant or conflicting information in the to-be-combined evidence groups, and accurately and efficiently conduct evidence combination and target recognition.