Symmetric Renyi-Permutation divergence and conflict management for random permutation set

Abstract

Recently, a theory called random permutation set (RPS) is proposed, which is the extension of Dempster–Shafer evidence theory. It defines a permutation event space (PES) to represent the order of events and uses the RPS to represent the support of permutation events. In addition, this theory fuses multiple RPSs by developing the left orthogonal sum (LOS) and right orthogonal sum (ROS) algorithms. However, our experiments have found that when two RPSs are highly conflicting, using the LOS or the ROS algorithm to fuse them will generate conflicting results. Therefore, this paper proposes the symmetric Renyi-Permutation (SRP) divergence of the RPS and a LOS-SRP algorithm for conflict management. Furthermore, existing RPS-based reliability assessment methods have the drawbacks of subjectivity and low accuracy, which prompts us to combine SRP divergence with the RPS and propose the RASRP algorithm for reliability assessment. Experimental results show that the RASRP algorithm performs better than the most advanced existing methods.