Stockwell Transform of Quadrature Stator Current-Based Fault Assessment in Induction Machine at Different Load Condition

Introduction: Conventional sensor-based fault detection techniques lead to huge manpower, high cost and requires a greater number of sensors. To overcome these flaws, current signature-based fault estimation is proposed. This method does not require any additional current sensors and is done with existed sensors employed for motor speed control. Further, the stator current is divided into two phase currents using Parks vector approach to get better resolution. The higher order noise due to data acquisition, EMI and other sources is removed using wavelet de-noising and the resolution on current spectrum is improved using advanced spectral analysis namely Stockwell Transform (ST). All the faults faced by induction motor like bearing damage, broken rotor, and stator inter turn faults are tested at different load conditions. The severity of the fault is identified using feature parameters like RMS and Standard Deviation of stator current after proposed signal processing. A 1.5 KW, 440 V and 50 Hz induction motor is used for expermental verification of proposed fault assessment criteria. Purpose: The present work attempts a novel approch to estimate incipient faults in Induction motor using motor current signature analysis (MCSA). Methods: Advanced spectral analysis namely Stockwell Transform (ST) along with Park’s vector approch technique has been adopted to diagnosis the various faults in induction motor (IM) using MCSA. Result and Conclusions: The proposed fault indexing parameters (FIP) with selected features like RMS and SD is successful in estimating the fault severity and its behaviour. Various faults at different load conditions are tested and concluded that, 3-hole rotor fault, 3-hole inner race fault and combined fault have high impact over other faults.