Fault diagnosis of rotating machinery based on evidence theory of evidence entropy

In order to improve the accuracy, Dempster-Shafer theory can be applied to the fault diagnosis of rotating machinery. However, the significance level of evidences is different actually when using evidence theory to fuse multi-symptom domains during fault diagnosis of rotating machinery. This paper presents evidence entropy to estimate significance level of evidence, i.e. the weight of evidences. Then, the evidences are adjusted according to the different weights, and the adjusted evidences are fused by the Dempster-Shafer combination rule. After that, the diagnosis result is obtained. Finally, through the real example, the research result shows that this method can be used to estimate significance level of evidence and reduces conflicting degree among evidences. Moreover, the effectiveness of the proposed method is also demonstrated.