A novel self-adaptive option method for sensitive failure component signals and its application in rolling bearings

Abstract

Bearing is the most vulnerable key part in rotating machine and bears important influence on the safety of equipment. Weakness and complexity are the two features of fault characteristic information carried by signals in the early stage of fault. For that, a fault is difficult to be recognized correctly. To identify a compound failure of bearing, the paper has brought forward a new self-adaptive option method for component signals that are sensitive to failure feature information of bearing. The sensitivity of kurtosis to bearing failure is exploited and the influence of signal complexity on the extraction of failure feature information is taken seriously, the paper has proposed the self-adaptive option method for component signals that are sensitive to failure feature by combined kurtosis with Complexity parameter included in Hjorth parameters. Furthermore, as the mid-value represents the general level of signal and is not affected by larger or smaller data, with the mid-values of kurtosis and Complexity parameter as the boundary, the paper has chosen the component signals which can more comprehensively show the failure features of bearing. Additionally, by principal component analysis (PCA), component signals selected are blended and reconstructed. Finally, by the Hilbert envelope spectrum of signals reconstructed, failure types of bearing are identified. To verify the effectiveness of presented method, the presented method is compared with conventional method on the basis of the exactly consistent data. The result indicates that the proposed method is superior to the traditional one in extracting fault information and identifying the multiple failure types of bearing.