Response of electrical drives to gear and bearing faults — Diagnosis under transient and steady state conditions

Abstract

Bearing and gear faults in systems using electrical drives share many commonalities and have some differences in they way that they can be detected. They both cause a mechanical impulse at a frequency that is related to the speed of the rotor. For neither is it possible to develop a state-space model relating the fault and its location and severity to the measured outputs. The measurements therefore require extensive processing to extract features that are related to a fault and to categorize it. Measurements physically close to the fault location (e.g. vibrations) are generally more useful to accurately determine this fault than measurements away from it (e.g. currents and voltage at the drive). Models of faults include fatigue (e.g. Paris model) and degradation due to bearing currents. In the recent past the techniques to identify these faults have been refined, tested extensively, and compared. They typically include signal conditioning, feature extraction (in the time and time-frequency domain) and categorization, which includes fault identification. Failure prognosis and use of multiple sensors are possible future directions of research to produce reliable estimated of condition and facilitate health management.