An Evidence Combination Rule Based on Transferable Belief Model and Application in Reliability Assessment With Multi-Source Data

Abstract

Increasingly complex system environments have put forward higher requirements for information fusion. Under this background, the application prospects of evidence theory which can function well in uncertain information representation and information fusion is promising. The effect of evidence theory on information fusion is closely related to evidence distance measurement method and evidence combination rules. Considering the difference of evidence information in belief level and that in decision level, this paper redefines the evidence distance and puts forward the measurement method of evidence credibility, thus realizing the correction of the original basic probability assignment function (BPA). Based on the evidence credibility, the evidence weight is generated, which leads to an efficient distribution of evidence conflict in the process of evidence combination. So, a new evidence combination rule is generated. According to a series of examples, the evidence distance proposed in this paper can effectively measure the evidence differences, and the new combination rules can reasonably distribute evidence conflicts thus avoiding evidence paradox. At the end of this paper, the new evidence distance and combination rule, combing with the Transferable Belief Model (TBM), are used to fuse the multi-source reliability test information, then the life cycle evaluation of a product is obtained. In addition, through the comparative analysis of the results, the feasibility and validity of this method being applied to practice are verified.

Keywords

• Evidence combination rule
• evidence paradox
• transferable belief model
• evidence distance
• reliability assessment