Vibration feature analysis for rolling element bearing diagnosis using Bayesian neural networks
Zhang, Sheng, Ma, Lin, & Mathew, Joseph (2005) Vibration feature analysis for rolling element bearing diagnosis using Bayesian neural networks. In Leong, Salman (Ed.) Asia Pacific Vibration Conference 2005, 22-25 November 2005, Langkawi Malaysia.
Condition monitoring in some form, is becoming an almost routine activity in many industries and is a key technology for machinery maintenance. Of the various techniques used, vibration monitoring is arguably the most popular given its direct ability to detect and diagnose faults in machinery. The vibration monitoring procedure comprises data acquisition; signal processing; condition assessment; and decision making. Extracting effective features plays a critical role in this procedure, since these features can indicate changes in machinery condition and track fault progression. Various features in both time and frequency domains can be extracted from the basic data by applying statistical analysis and signal processing. The sensitivities of the features have however not been compared. It is now well accepted that using multiple features can enhance condition assessment. It would be best if an optimal feature combination can be selected. This research examines feature analysis issues using a SpectraQuest machinery fault simulator. Vibration signals were collected from rolling element bearings under four conditions, healthy, inner race fault, outer race fault and ball fault. Sixteen vibration features extracted from both time and frequency domains were investigated for bearing diagnosis. The individual features were examined against their separation capabilities using discrimination measures. A Bayesian Neural Network (BNN) was then used to automatically determine the relevance of each feature. This research clarifies the sensitivity of individual vibration features and provides an approach for selecting optimal feature combination for rolling element bearing fault diagnosis.
Citation countsare sourced monthly fromand citation databases.
These databases contain citations from different subsets of available publications and different time periods and thus the citation count from each is usually different. Some works are not in either database and no count is displayed. Scopus includes citations from articles published in 1996 onwards, and Web of Science® generally from 1980 onwards.
Citations counts from theindexing service can be viewed at the linked Google Scholar™ search.
|Item Type:||Conference Paper|
|Additional Information:||For more information please contact the author: firstname.lastname@example.org|
|Keywords:||Vibration analysis, rolling element bearing, Bayesian neural networks|
|Subjects:||Australian and New Zealand Standard Research Classification > ENGINEERING (090000) > MECHANICAL ENGINEERING (091300) > Mechanical Engineering not elsewhere classified (091399)|
Australian and New Zealand Standard Research Classification > ENGINEERING (090000) > INTERDISCIPLINARY ENGINEERING (091500) > Interdisciplinary Engineering not elsewhere classified (091599)
|Divisions:||Past > QUT Faculties & Divisions > Faculty of Built Environment and Engineering|
|Copyright Owner:||Copyright 2005 (please consult author)|
|Deposited On:||01 Dec 2006|
|Last Modified:||29 Feb 2012 23:12|
Repository Staff Only: item control page