Performance analysis of wavelet based techniques for electrical faults signature extraction for squirrel cage induction motor

Induction motors share a domineering role in engineering applications in industries. Even though they are highly reliable, they are prone to assorted faults. Electrical faults such as unbalanced supply and rotor bar breakage contribute significantly to these faults and imperfections. Detection of the faults in embryonic stages is key to timely scheduled maintenance. Motor Current Signature Analysis (MCSA) is a typical tool for fault detection in motors at constant torque loads. But pulsating load and variable load torque operations put down arduous and tough constraints on part of resolution. Multi Resolution Analysis (MRA) is an approach to trounce the impediments of frequency resolution. Discrete Wavelet Transform (DWT) and Wavelet Packet Transform (WPT) are two contemporary techniques in time-space domain. This paper presents comparative analysis of these two techniques for feature extraction of electrical faults in Induction motors. Instead of analysing the three phase currents of the motor independently, the direct (i_d) current component is made use of. The experimentation has been performed on a 3-phase, 1.5kW, 4P, 1440 RPM squirrel cage induction motor. Fault Signature Extraction (FSE) is carried out by applying signal energy difference evaluation algorithm on DWT and WPT coefficients for various many cases of unbalanced supply and rotor bar breakage faults. Analysis begets to establish that both the above mentioned techniques attest to be significant for the purpose of fault detection. To classify the fault types, other high-ended techniques need to be associated and analysed.