Single-turn fault detection in induction machine using complex-wavelet-based method

Interturn short circuit is often confused with voltage imbalance in induction machines. Therefore, detection and classification of single-turn fault (TF) are becoming important in the presence of voltage imbalances, under various loading conditions. Substantial studies are conducted on the interturn fault detection, but a comprehensive method for classifying the faults at different operating points of the machine, under varying supply conditions, is still a challenge. This is a critical problem in industries since the induction motors form the major workhorses. The artificial-intelligence-based techniques are advanced methods in fault monitoring. This, when combined with optimization techniques, is expected to give improved and accurate results with minimum false alarms. In this paper, a technique is developed, based on recent developments in the wavelet-based analysis, particularly in the complex wavelet domain. The support vector machines are adopted for comparing the classification accuracy obtained using complex-wavelet- and standard discrete-wavelet-based methods. The receiver operating characteristic curves indicate that the fault detection, down to single turn, is feasible using a single current sensor