A Bayesian Optimized Multi Harmonics Index Approach for Fault Diagnosis of Gearboxes

The project aims to design and develop an algorithm that can diagnose faults in planetary and fixed-axis gearboxes. Common faults in gearboxes are often masked by the Gear Meshing Frequency, making them difficult to detect via classical spectral analysis. A multi-harmonic index approach based on an adaptive combined envelope spectrum is selected as the foundational algorithm due to its robustness to noise in bearing analysis and its cyclostationary capabilities. Moreover, a Bayesian algorithm is proposed to select the ideal frequency band, improving the time complexity as compared to other optimization algorithms.

Faculty Supervisor:

Xihui (Larry) Liang

Student:

Partner:

University of Technology Sydney

Discipline:

Engineering

Sector:

Education

University:

University of Manitoba

Program: