A Novel Data-Driven Feature Extraction Strategy and Its Application in Looseness Detection of Rotor-Bearing System

Yulai Zhao, Junzhe Lin, Xiaowei Wang, Qingkai Han () and Yang Liu
Additional contact information
Yulai Zhao: School of Mechanical Engineering and Automation, Northeastern University, Shenyang 110819, China
Junzhe Lin: School of Mechanical Engineering and Automation, Northeastern University, Shenyang 110819, China
Xiaowei Wang: School of Mechanical Engineering and Automation, Northeastern University, Shenyang 110819, China
Qingkai Han: School of Mechanical Engineering and Automation, Northeastern University, Shenyang 110819, China
Yang Liu: School of Mechanical Engineering and Automation, Northeastern University, Shenyang 110819, China

Mathematics, 2023, vol. 11, issue 12, 1-18

Abstract: During the service of rotating machinery, the pedestal bolts are prone to looseness due to the vibration environment, which affects the performance of rotating machinery and generate potential safety hazard. To monitor the occurrence and deterioration of the pedestal looseness in time, this paper proposes a data-driven diagnosis strategy for the rotor-bearing system. Firstly, the finite element model of a rotor-bearing system is established, which considers the piecewise nonlinear force caused by pedestal looseness. Taking the vibration response as output and periodic unbalanced force as input, the system’s NARX (Nonlinear Auto-Regressive with exogenous inputs) model is identified. Then GALEs (Generalized Associated Linear Equations) are used to evaluate NOFRFs (Nonlinear Output Frequency Response Functions) of the NARX model. Based on the first three-order NOFRFs, the analytical expression of the second-order optimal weighted contribution rate is derived and used as a new health indicator. The simulation results show that compared with the conventional NOFRFs-based health indicator, the new indicator is more sensitive to weak looseness. Finally, a rotor-bearing test rig was built, and the pedestal looseness was performed. The experiment results show that as the degree of looseness increases, the new indicator SRm shows a monotonous upward trend, increasing from 0.48 in no looseness to 0.90 in severe looseness, a rise of 89.7%. However, the traditional indicator Fe2 has no monotonicity, which further verifies the sensitivity of the first three-order NOFRFs-based second-order optimal weighted contribution rate and the effectiveness of the proposed data-driven feature extraction strategy.

Keywords: pedestal looseness; data-driven dynamic; feature extraction; NARX; nonlinear output frequency response functions (search for similar items in EconPapers)
JEL-codes: C (search for similar items in EconPapers)
Date: 2023
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
https://www.mdpi.com/2227-7390/11/12/2769/pdf (application/pdf)
https://www.mdpi.com/2227-7390/11/12/2769/ (text/html)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:gam:jmathe:v:11:y:2023:i:12:p:2769-:d:1174591

Access Statistics for this article

Mathematics is currently edited by Ms. Emma He

More articles in Mathematics from MDPI
Bibliographic data for series maintained by MDPI Indexing Manager ().