An Advanced Weighted Evidence Combination Method for Multisensor Data Fusion in IoT

Abstract

Dempster-Shafer theory is an essential tool for modelling and reasoning under uncertainty, and it is an effective approach for multisensor data fusion. It is extensively deployed in many fields such as fault diagnosis, image processing, pattern recognition, etc. However, Dempster’s combination rule is often subject to counter-intuitive results when the sources highly conflict; several methods have been proposed in the literature to solve this problem. In this paper, a weighted evidence combination method based on evidence distance, evidence angle, and information volume is proposed to overcome the shortcomings of the classical Dempster’s combination rule. To investigate the effectiveness and performance of the proposed method, a comparative study with different state of the art methods using both benchmark numerical example and Fault diagnosis application has been carried out.