BRIEF OVERVIEW OF NEW RESULTS OF THE DOCTORAL THESIS

BRIEF OVERVIEW OF NEW RESULTS OF THE DOCTORAL THESIS
(Information to be posted on the Website)

Topic of the thesis: Fault diagnosis of gearboxes using time-frequency analysis of
mechanical vibrations
Specialization: Engineering Mechanics – Code No: 62520101
Name of PhD. Student: Nguyen Trong Du
Supervisor: Assoc. Prof. Dr
.
Nguyen Phong Dien
Training Institution: Hanoi University of Science and Technology

Brief overview of the new results of the thesis
On the theoretical aspect:
- Systematic presentation of the mathematical background of some linear time-
frequency transforms such as Windowed Fourier Transform (WFT), Continuous Wavelet
Transform (CWT) and the Wavelet-based Synchrosqueezing Transform (WSST);
development of the corresponding algorithms and a computer program in MATLAB for
these transforms.
- Propose a solution to increase the resolution of the time-frequency distribution of a
vibration signal using the WSST.
- Propose an integrated approach between Wavelet Packet Transform and Artificial
Neural Network to automatically detect and classify some type of faults of rotating elements
in a gearbox.

In terms of applications:
- Successful application of new and modern signal processing methods such as
Generalized Synchrosqueezing Transform (GST) to increase the diagnostic accuracy.
- Suggestion and application of an improved method of time-synchronous averaging
(TSA) for fault detection in multi-stage gearboxes using only one tachometer or without
tachometer. The proposed TSA-method is significant in reducing the cost of equipment for

gear diagnostic systems.
- Suggestion and application of a diagnostic procedure based on time-frequency analysis
of mechanical vibrations to detect and localize faults in helical gearboxes operating under
fluctuating load conditions and varying rotational speeds. This procedure has been
demonstrated and verified by a gearbox test-rig and in-situ experiments.
- A self-written signal processing toolbox in MATLAB that integrates a number of
digital signal analysis methods in time-frequency domain has been successfully applied on
experimental studies. It can be shown that the toolbox is capable of applying in practice for
early detection and localization of local faults in the rotating parts of the gearbox.
Hanoi, June 30, 2015

Supervisor
(Signature)




Assoc. Prof. Dr. Nguyen Phong Dien
PhD. Student
(Signature)




Nguyen Trong Du