Dynamic health index extraction for incipient bearing degradation detection. (September 2022)

• Record Type: Journal Article
• Authors: Ye, Xinlai; Li, Guoyan; Meng, Linghui; Lu, Guoliang
• Is Part Of: ISA transactions. Volume 128(2022)Part B
• Page Start: 535
• Page End: 549
• Publication Date: 2022-09
• Subjects: Dynamic modeling -- Graph model -- Dynamic graph analysis -- Degradation detection -- Roller element bearing; Engineering instruments -- Periodicals; Engineering instruments; Periodicals; Electronic journals; 629.805
• DOI: 10.1016/j.isatra.2021.11.029
• Languages: English
• ISSNs: 0019-0578

Abstract

Abstract: Performance degradation is a natural phenomenon for mechanical roller element bearings (REBs) during their long-term service time. It is essential to extract an effective dynamic health index, that can describe and quantify the dynamic characteristics of REBs health status, for automated detection of REB degradation at an early stage. This study presents a new numerical computation method to achieve this end, which can consider and utilize useful information from different individual indices. First, graph-based modeling integrated with dynamic analysis is performed on each channel of individual indices to solve the non-stationary and noise problems. The adaptive inputs weighting (AIW) fusion technique is adopted to assign adaptive weights to each graph-enhanced channel for the purpose of multi-channel graph information fusion. The resulting comprehensive index is finally fed to a commonly-used hypothesis test for decision making. Comprehensive evaluations conducted on simulation and real scenarios demonstrated the significant improvements of the proposed method and its great potential in practical applications. Highlights: This paper focuses on health index extraction for incipient bearing degradation detection. A new graph-based method for individual index enhancement and fusion is proposed. A new comprehensive health index is created based on multi-channel graph information fusion. The accuracy and noise resistance on incipient bearing degradation detection are improved.