An Improved Multisensor Data Fusion Method and Its Application in Fault Diagnosis

Abstract

A multisensor system is used to improve the ability of monitoring and tracking accuracy of an engineering system since only one sensor is not enough, whereupon, faulty diagnosis, as a typical information fusion problem which is under multisensor environment, has attracted much attention in recent years. The evidence theory has been widely used to solve this problem. However, when there is a high level of conflict between the information gathered from different sensors, the counter-intuitive result could be obtained when using the classical Dempster’s combination rule. To address this problem, an improved multisensor data fusion method is proposed to fuse the data collected from multisensors. Some numerical examples are illustrated to show that the proposed method is effective and feasible. Moreover, the fusion results using different methods are analyzed, which indicate the superiority and stronger application of the proposed method in the field of fault diagnosis.

Keywords

• Conflict evidence
• Dempster’s combination rule
• belief entropy
• Dempster-Shafer evidence theory
• information theory