Combination method of conflict evidences based on evidence similarity

Wenhao Bi, An Zhang, Yuan Yuan

Research output: Contribution to journalArticlepeer-review

25 Scopus citations

Abstract

Aiming at the problem that the traditional Dempster-Shafer (D-S) evidence theory cannot deal with conflicted evidences effectively and correctly, this paper points out that the key issue of this problem is to measure the degree of conflict between evidences correctly after analyzing various improved methods. The existing evidence conflict measure methods are analyzed, and a new evidence conflict measure method called evidence similarity measure based on the Tanimoto measure is proposed, while a new evidence combination method is proposed on the basis of evidence similarity measure. Firstly, the conflict degrees between evidences are obtained through the evidence similarity measure. Then the evidence sources are modified based on the credibility of different evidences and the weights of conflicted parts of evidences on different focal elements are determined. Finally, the fusion result is obtained by this method. Numerical examples show that the proposed method can effectively fuse evidences when evidences are consistent or highly conflicted, and it has a fast convergence speed, a high degree of accuracy and good adaptability.

Original languageEnglish
Article number7978023
Pages (from-to)503-513
Number of pages11
JournalJournal of Systems Engineering and Electronics
Volume28
Issue number3
DOIs
StatePublished - 2017

Keywords

  • conflict measure
  • evidence conflict
  • evidence similarity
  • evidence theory
  • information fusion

Fingerprint

Dive into the research topics of 'Combination method of conflict evidences based on evidence similarity'. Together they form a unique fingerprint.

Cite this