Detection of broken rotor bar fault in induction motor at various load conditions using wavelet transforms

Early detection of incipient rotor bar faults are important for efficient operation of large induction motors. This paper presents a methodology to detect rotor bar faults under no-load and loaded conditions, using wavelet transforms. In wavelet analysis, the wavelet coefficients are dependent on the basis function chosen. If these coefficients are used to capture the signature of a fault, then it is desirable to select a wavelet that produces the best results for the signal being analyzed. This paper presents the results obtained from a comparative study done using different discrete wavelet transforms for detection of broken rotor bar fault. Results indicate that the energy content of the wavelet coefficients are sufficient to distinguish between a healthy and faulty machine. The simulation of the proposed methodology is carried out using MATLAB /SIMULINK.