an intelligent system for fault diagnosis in gearboxes
abstract
gearboxes are commonly used in rotating machinery for power transmission. a gearbox consists
of shafts, gears, and bearings, each component having specific mechanical dynamics and fault
properties. reliable gearbox fault detection and health monitoring techniques are critically
needed in industries for more efficient predictive maintenance applications. the objective of this
work is to develop a new technology for health monitoring of gearboxes. firstly, a new wavelet
analysis method is technique for analysis of gear faults in a gearbox with demodulation from
other rotating components such as shaft and bearings. secondly, a mode decomposition
technique is proposed to highlight bearing fault features in a gearbox. thirdly, a new evolving
neuro-fuzzy (enf) classifier is developed to integrate the merits of different fault detection
techniques for real-time health condition monitoring of gear systems. the effectiveness of the
proposed techniques is verified by simulation and experimental tests.