Graduation date: 2006
Date presented: 2005-08-26
Advisors: Annette von Jouanne, Alan K. Wallace.
Committee members: Joseph Zaworski, Thomas Plant
The detection of motor faults at their incipient stage is of prime importance to any
industrial plant. The introduction of adjustable speed drives has improved the control and
the efficiency of induction motors, however, this has changed the nature of motor faults
and how they can be detected.
Current signature analysis has caught the attention of researchers as a mature and
simple technique for motor fault diagnosis. In this research three main ways of analyzing
the current signature for fault detection have been investigated. These are: the power
spectral density analysis, the current negative- and positive-sequence components, and
the Park’s vector approach.
Three major induction motor faults have been experimentally tested for the above
diagnosis techniques: the bearing fault, the broken rotor bar, and the air gap dynamic
eccentricity. Using an adjustable speed drive for controlling the motor while applying
these fault detection techniques has been compared to the supply of the motor directly
from the “mains” source and to a pure sinusoidal supply through a programmable source.
This research has proved that using the power spectral density analysis is a good
tool for induction motor fault detection regardless of the source of supply. This technique
can be easily implemented in standard commercial adjustable speed drives, with no
additional hardware requirements.