Detailed investigation of negative sequence current compensation technique for stator shorted turn fault detection of induction motors.

Online condition monitoring is preferred specifically in critical induction machines for timely analysis of their health and to predict any incipient failures. Negative sequence current monitoring is one of the well-known condition monitoring applications that capable of identifying stator faults in induction motor as well as producing fast and reliable results using a simple measurement technique. However, the technique is sensitive to asymmetrical sources of negative sequence, not only due to motor faults but also due to the disturbances such as sensor calibration, inherent non-idealities (such as inherent machine asymmetries), external effects (such as load change and supply voltage unbalances). This thesis addresses the limitations in shorted turn fault detection while considering real system disturbances and non-linearities using a dynamic motor model which is also verified by the experimental studies. The thesis provides a comprehensive investigation of various negative sequence contributors such as measurement asymmetry, inherent asymmetry, voltage unbalance, shorted turn faults, eccentricity and broken rotor bar faults under load variations, and presents practical results. To allow accurate shorted turn fault detection, this thesis also provides an effective phasor compensation technique for these disturbances. A new detailed compensation by demonstrating step-by-step disturbance extraction to obtain the real negative sequence current due to shorted turn faults are also presented. It is demonstrated that this technique allows even a single stator shorted turn faults to be detected as well as successfully eliminating the effect of inherent asymmetry and voltage unbalances under motor load variations.Thesis (Ph.D.) -- University of Adelaide, School of Electrical and Electronic Engineering, 201