| dc.description.abstract | Nowadays, modern rotating machinery industries such as automotive, aerospace, turbo
machinery, chemical plants, and power generation stations are rapidly increasing in complexity
and in their everyday operations, which demand their systems to operate at higher reliability,
extreme safety, and with lower production and maintenance costs. Therefore, accurate fault
diagnosis of machine failure is vital to the operation of the related industries. The majority of
machine imperfections has been related to gearbox faults (e.g., gears, shafts and bearings), which
are subject to damage modes such as fatigue, impacts, and overloading. Faults not detected in
time can result in severe damage to machinery, catastrophic injuries, and substantial financial
losses. On the other hand, if a fault is detected in its early stages, corrective and preventive action
can be taken to avoid any significant machine failure. Vibration monitoring, a method that is
widely used to determine the condition of various mechanical systems, will be applied in this
work. In data acquisition, a transducer is attached to the structure under investigation and the
vibration signal is recorded. This signal is then processed to extract representative features for
fault detection. Signal processing techniques are therefore required to extract representative
features to assess the health condition of gearbox components. However, in practice, the
theoretical frequencies and characteristic features of gearbox faults may be modulated and
masked by parasitical frequencies due to numerous noisy vibrations, as well as by the complexity
of the transmission mechanics. To solve the related problems, the objective of this research work
is to propose new signal processing technologies to evaluate gearbox health conditions. This
work will focus on fixed-axis gearboxes, in which all gears are designed to rotate around their
perspective fixed centers. Firstly, an enhanced morphological filtering (eM) technique is
proposed to improve signal-to-noise ratio. Secondly, under controlled operating conditions, an
integrated Hilbert Huang transform (iHT) method is suggested for bearing fault detection.
Thirdly, a leakage-free resonance sparse decomposition (LRSD)-based technique is developed
for advanced vibration signal analysis to eliminate random noise and to recognize characteristic
features for bearing in gearboxes health conditions. The effectiveness of the proposed techniques
is verified by a series of experimental tests corresponding to different bearing and gearbox
conditions. | en_US |
