Prediction of rotor slot size and rotor fault in squirrel cage induction Motor using ridge regression

Abstract

Squirrel Cage Induction Motor (SCIM) is widely used in various industries such as cement, textiles, oil, gas, waste water treatment, mining and water pumps. Continuous monitoring of Average Rotor Slot Width Variation (ARSWV) in SCIM is predicted using Wavelet Transform and Regression such as Short Time Fourier Transform- Multiple Linear Regression (STFT-MLR) and Transverse Dyadic Wavelet Transform-Ridge Regression (TDyWT-RR) during the running condition of the SCIM. The variation of the rotor slot width in SCIM is due to high thermal stress and magnetic flux. In the proposed approach, Symmetrized Dot Pattern (SDP), Scale Invariant Feature Transform (SIFT) and Least Square-Support Vector Machine (LS-SVM) are used to identify the rotor faults in SCIM. Multimodal sensor signals such as vibration, temperature and Gaint Magnetoresistance (GMR) are acquired from SCIM, converted into 2D images through SIFT, and images are obtained for induced faults in the SCIM. Manual measurement of ARSWV is performed through Microscopic Camera (MC). ARSWV >3.5 % damages the rotor, which is experimentally verified from SCIM. The proposed ARSWV method prediction accuracy is 96.7 %, when compared to microscopic camera image based ARSWV measurement. The rotor fault detection using SDP, SIFT and LS-SVM is about 98 % when compared to traditional method.