Work in progress: Multi-dimensionless parameters fusion method based on improved D-S evidence theory

Abstract

Aiming at the invalidation problem of Dempster-Shafer evidence combination rule with high conflict, a fault diagnosis method of multi-dimensionless parameters information fusion based on novel D-S algorithm improved by the weighted concept is proposed. Firstly, prior weights are obtained by the basis of inherent reliability of multi-dimensionless parameters diagnosis information, and a similarity matrix between evidences is constructed as posterior weights. Then the prior weights and posterior weights are combined regarded as the weight coefficient of the evidence and evidences are weighted with these weight coefficients. Finally the D-S evidence theory is used to fuse the weighted evidences and the final diagnosis result can be obtained. The experimental results demonstrated that, compared with other methods, the new algorithm can improve accuracy of fault diagnosis and reduce influence of conflict evidence effectively.