Fault detection and severity assessment for gearboxes

The gearbox is a classical mechanical unit and has been widely used in modern power transmission systems. Fault detection and severity assessment of gearboxes prior to their failure can prevent the sudden failure of gearboxes, as well as enable condition-based maintenance and thus reduce maintenance costs. However, the fault detection and severity assessment are challenging when the gearbox operates under time-varying rotational speed (TVRS) conditions. Unlike the constant speed condition, the condition monitoring data, e.g., vibration, becomes non-stationary due to amplitude and frequency modulations induced from TVRS. Conventional signal processing and analysis tools are no longer applicable. Advanced signal analytic algorithms are in demand. This project aims at proposing an advanced statistical time series model to globally represent the gearbox vibration signal under TVRS. Fault detection and severity assessment objectives are subsequently achieved. The hypothesis is that the time series model-based method will have an improved fault detection rate and severity assessment accuracy compared with existing methods. The result of this project will fundamentally benefit the safety, maintenance and operation of power transmission systems.

Faculty Supervisor:

Ming Zuo

Student:

Partner:

Georgia Institute of Technology;University of New South Wales

Discipline:

Engineering

Sector:

Education

University:

University of Alberta

Program: