Multi-Fault Diagnosis of Induction Motors based on Adaptive Wavelet Packet Transform

Despite their inherent robustness and cost-effectiveness, various faults occur in Induction Motors (IM). Thus, motivations for Condition Monitoring (CM) of IM are increasingly growing. The vast majority of CM works focus on monitoring isolated faults in bearings, rotor or in stator. Nevertheless, this type of CM can mislead the complete evaluation of the IM state, since they could be simultaneously affected by multiple combined faults. This paper proposes a low cost sensors-based CM using the motor current analysis for the diagnosis of several electrical and mechanical faults in IM both separated and combined. A novel signal-processing tool, the Adaptive Wavelet Packet Transform (AWPT), processes the data acquired from the current sensors. Several bearing faults and broken rotor bars that include the incipient or partially broken bar fault are diagnosed under various operating loads. A high fault detection accuracy and an important gain in implementation cost prove the competitive performance of the proposed approach.