The trend toward ever-expanding variable speed induction motor applications results in the need for reliable condition monitoring and detection schemes for closed-loop motor-drive systems. This research focuses on the detection of air gap eccentricity in induction motors supplied by a vector-controlled drive. The majority of existing eccentricity detection techniques is based on monitoring fault harmonics in the stator current. This research analyzes the distribution of the eccentricity-related fault harmonics between the stator voltage and current, and points out that monitoring only the stator current is insufficient. When the motor is supplied by a vector-controlled drive, both voltage and current become modulated signals and contain fault harmonics. Either stator voltage or current can contain larger fault harmonics due to the influence of the drive controllers and the mechanical load. Therefore, a combination of monitoring both variables is necessary to ensure good detection reliability. Furthermore, with an AC drive, the motor speed and load will change widely, which changes fault harmonics too. A new detection scheme using an artificial neural network is proposed to incorporate the influence of changing operating conditions into the fault detection. This detection scheme is more reliable and cost efficient. In addition, a new off-line non-invasive eccentricity detection method is proposed by using the surge test. Simulation and experiments are conducted to validate the feasibilities of the proposed detection schemes.

This thesis describes properties of windings of electric rotating machines and their insulation systems. There are winding and insulation low voltage machines tests listed with their procedures and criteria. Further it deals with coordination methodology and the last part contains execution and results assessment of tests conducted on stator samples.