An ensemble deep autoencoders based on asymmetric structure for operational reliability assessment of bearings. (October 2022)

• Record Type: Journal Article
• Authors: Jin, Cheng; Yang, Xingyu; Ma, Hongbo; Wu, Xiaodong; Yang, Guanbin
• Is Part Of: Advances in mechanical engineering. Volume 14:Number 10(2022)
• Publication Date: 2022-10
• Subjects: Operational reliability assessment -- rolling bearing -- ensemble deep autoencoder -- asymmetric structure -- unsupervised feature extraction; Mechanical engineering -- Periodicals; 621.05
• DOI: 10.1177/16878132221130573
• Languages: English
• ISSNs: 1687-8132

Abstract

At present, with the rapid growth of manufacturing and big data, reliability technology has gradually become a topical issue in the industrial field. Aiming at the operation reliability assessment of rolling bearings, this paper proposes a bearings operational reliability assessment using an ensemble deep autoencoder based on asymmetric structure. In this method, an ensemble deep autoencoder is used to adaptively learn degradation features from condition monitoring data, where the ensemble deep autoencoder adopts an asymmetric structure with different activation functions in the encoder and decoder. Then, the learned features are classified by correlation analysis, and the typical features in each category are selected. Finally, the operation reliability of rolling bearings is evaluated through the definition of reliability based on Mahalanobis distance. Through the example evaluation of rolling bearing operation reliability and comparison with other feature extraction methods, it can be concluded that this method has stronger feature extraction ability and can effectively show the trend of bearing degradation.