Vibration Signal Analysis for Planetary Gearbox Fault Diagnosis

• Author / Creator

  Liu, Libin

• Vibration signal analysis has been widely used for planetary gearbox condition monitoring and fault diagnosis. Vibration signals are normally measured by sensors mounted on a planetary gearbox’s housing. A vibration signal contains rich information about the health condition of the machinery. However, the vibration characteristics of a planetary gearbox are quite complicated because of the complex structure and kinematics. In a planetary gearbox, multiple sun-planet gear pairs and multiple ring-planet gear pairs are in mesh simultaneously, introducing multiple vibration sources from the sun gear, the ring gear, and the planet gears. In addition, due to carrier rotation, the multiple vibration sources are subject to different transmission path effects. Multiple vibration sources and the different time-varying transmission path effects lead to challenges with understanding vibration characteristics and diagnosing faults in a planetary gearbox by vibration-based methods.
  The aim of this thesis study is to develop effective vibration signal analysis methods for planetary gearbox fault diagnosis. Four research topics are addressed. Firstly, a vibration signal model with transmission path effects is developed to characterize the properties of planetary gearbox vibration signals. Using the model, realistic vibration signals are simulated to aid the development of vibration signal analysis methods for planetary gearbox fault diagnosis in the three remaining research topics. As the second research topic, a spectrogram-free copula-based Time-Frequency Distribution (TFD) construction method is developed for energy density representation of a one-dimensional vibration signal with properties of being positive and free from cross-term interference, with correct energy marginals and high time-frequency resolution. It is applied to planetary gearbox fault diagnosis in a simulated case study and an experimental case study. The results show that, being free from the Heisenberg uncertainty principle and with high time-frequency resolution, the identification of fault-related frequencies with fine frequency resolution and the location of fault-induced impulses with fine time resolution can be achieved simultaneously. As the third research topic, a method of fault feature extraction via dimension reduction on the time-frequency energy density of a one-dimensional vibration signal is explored and developed by Non-Negative Matrix Factorization (NNMF). The spectrogram-free copula-based TFD constructed in the second research topic serves as the time-frequency energy density. Validation is performed through a simulated case study and an experimental case study. Inspired by the dependence analysis involved in the second research topic and the third research topic, in the fourth research topic, a dependence-based feature vector is developed for planetary gearbox fault classification. The method is tested based on the dependence between the raw one-dimensional vibration signal and its Intrinsic Mode Functions (IMFs). The IMFs are obtained by the Ensemble Empirical Mode Decomposition (EEMD). The dependence is revealed to be an upper tail dependence described by the Gumbel-Hougaard (GH) copula. The proposed dependence-based feature vector is developed through simulated vibration signal analysis and defined as the pair of GH copula coefficients regarding the first two IMFs. Validation is conducted through an experimental case study.
  The thesis study would promote the state of the art of research on vibration signal analysis for planetary gearbox fault diagnosis. Knowledge generated from the four research topics will provide practical engineers with powerful tools for diagnosing faults in planetary gearboxes, thus benefiting industrial applications of planetary gearboxes, such as wind turbines and helicopters, with high reliability, safety, and low operation and maintenance cost.
  The planetary gearbox of interest in this thesis study is under stationary operation conditions with a single gear tooth fault. Further analysis on cases under non-stationary operation conditions and/or with multiple gear tooth faults will be studied in future work.

• Subjects / Keywords

  • Vibration Signal Model
  • Positive Time-Frequency Distribution
  • Dependence-based Feature Vector
  • Transmission Path Effect
  • Non-Negative Matrix Factorization
  • Vibration Signal Analysis
  • Copula Theory
  • Planetary Gearbox Fault Diagnosis

• Graduation date

  Fall 2018

• Type of Item

  Thesis

• Degree

  Doctor of Philosophy

• DOI

  https://doi.org/10.7939/R3862BT40

• License

  Permission is hereby granted to the University of Alberta Libraries to reproduce single copies of this thesis and to lend or sell such copies for private, scholarly or scientific research purposes only. Where the thesis is converted to, or otherwise made available in digital form, the University of Alberta will advise potential users of the thesis of these terms. The author reserves all other publication and other rights in association with the copyright in the thesis and, except as herein before provided, neither the thesis nor any substantial portion thereof may be printed or otherwise reproduced in any material form whatsoever without the author's prior written permission.