A novel evidence combination method based on stochastic approach for link-structure analysis algorithm and Lance-Williams distance

Abstract

In response to the traditional Dempster–Shafer (D-S) combination rule that cannot handle highly conflicting evidence, an evidence combination method based on the stochastic approach for link-structure analysis (SALSA) algorithm combined with Lance-Williams distance is proposed. Firstly, the degree of conflict between evidences is calculated based on the number of correlation coefficients between evidences. Then, the evidences with a number of correlation coefficients greater than the average number of correlation coefficients of evidence are connected to construct an evidence association network. The authority weight of the evidence is calculated based on the number of citations in the concept of SALSA algorithm combined with the support of the evidence. Subsequently, the Lance-Williams distance between the evidences is calculated and transformed into support of the evidence. Next, the authority weight and support of evidence are combined to jointly construct a novel correction coefficient to correct the evidence. Finally, the corrected evidence is fused using the D-S combination rule to obtain the final fusion result. The numerical results verify that the method proposed in this paper can effectively solve the problem of the traditional D-S combination rule being unable to handle highly conflicting evidence.

Keywords

• D-S combination rule
• SALSA algorithm
• Lance-Williams distance
• Correlation coefficient