Early Diagnosis of Induction Machine Stator Winding Faults by Discrete Wavelet Transform

Abstract

AbstractElectrical machines are useful in nuclear power plants, military applications, domestic appliances and industrial utilities. The relevant literature indicates that the induction machine takes about ninety percent of all electrical machine globally used in the industry. The risk of the failure of this machine can be avoided if the proper diagnostic scheme is designed and implemented to detect failure/impending faults at an incipient stage. This paper describes Discrete Wavelet Transform (DWT) based diagnosis technique that analyses stator currents of induction machine under healthy and faulty conditions. This enables the extraction of harmonic components caused by each signal for further analysis. On one hand, the maximum energy for each signal gives information about the time a certain fault occurs in the machine and the time it starts deviating from the normal (healthy) state. On the other hand, a Fault Index (FI) is generated for each situation in order to classify the state of the machine into normal, medium or high. The colours; green, yellow and red are used to represent the normal, medium and high states respectively. Should the faults at medium state are not detected on time, it may lead a more serious fault (high state). Thus, this technique is robust to identify a fault of an induction machine at the incipient state with a very minimal time.