Incipient turn fault detection and condition monitoring of induction machine using analytical wavelet transform

Abstract

Diagnosis and monitoring the condition of induction machines and supply system is critical for industries. Incipient fault detection has received reasonable attention in recent years. In this paper, a method based on complex wavelets is proposed for incipient fault detection and condition monitoring. A complex wavelet-support vector machine (SVM) classifier-based method is developed which takes into account four conditions: healthy, turn fault (TF) under balanced supply conditions, voltage imbalance, and interturn fault with voltage imbalance, both occurring at the same time. The performance metrics show the ability of the technique to identify the fault at an early stage and it also provides additional information regarding which of the four conditions is prevailing at a given time. Voltage imbalance and turn fault are often confused. Both affect the performance of the machine and the unbalanced voltage condition considerably reduces the winding insulation life due to overheating. This necessitates the precise identification of the supply condition along with the fault diagnosis. A comparison of the proposed method with standard discrete wavelet transform (DWT) shows its effectiveness in providing reliable information under variable supply-frequency conditions. The proposed technique is also tested in presence of high resistance connections (HRCs), which shows its isolating capability.