Estimation of Fan Bearing Degradation Using Acoustic Emission Analysis and Mahalanobis Distance

In order to estimate health conditions of fan bearings and predict their remaining life before failure, relevant features associated with their degradation must be identified. There are few published papers that deal with the selection of relevant features from acoustic emission signals for health condition estimation and life prediction of bearings in cooling fans. For this study, acoustic emission signals were measured periodically during stress tests of cooling fans, from which a total of fifteen different acoustic emission features were extracted. Correlation coefficients between the acoustic emission features and stress duration were calculated in order to identify features which were associated with bearing degradation. The distinguishing characteristics of acoustic emission features were also described. By using a health index based on a Mahalanobis distance integrating the acoustic emission features selected from the correlation analysis, the health conditions of fan bearings were estimated over the duration of the stress test. Failure analysis of asreceived and failed fan bearings was conducted to identify physical defects on surfaces of bearing elements and relate the evolution of the physical defects to the generation of acoustic emission. The approach presented in this paper helps to identify acoustic emission features associated with the evolution of the physical condition of fan bearings.