Detection of Broken Rotor Bars Using Multilevel Wavelet Decomposition

Abstract

Induction motors find widespread applications in the industrial sector. These motors are quite often subjected to hostile service conditions and heavy duty cycles which make them vulnerable to different kinds of mechanical faults. Therefore, non-invasive condition monitoring is very crucial in enabling safe operation, timely maintenance and desirable efficiency. Among the faults, the rotor broken bar condition requires careful study basically because the failure is silent and consequences catastrophic. Prompt detection and accurate diagnosis of such faults reduce the higher costs and increased downtimes associated with corrective maintenance. This work suggests an online fault detection technique based on multilevel wavelet decomposition. The wavelet coefficients provide an insight into the operational condition of the machine and is used as the feature to extract fault details.