A Boosting Classifier for Induction Motor Fault Diagnosis

Many fault diagnosis techniques have been proposed in literature for motor fault detection, however, each having its own merits and limitations. A new boosting classifier is developed in this paper to classify features from three information domains, i.e., time domain, frequency domain and time-frequency domain for fault diagnosis. In the proposed boosting classifier, a new noise regulation mechanism is proposed to address the noise samples, in order to derive more robust fault diagnosis. The effectiveness of the developed boosting classifier is verified by the experiments of induction motors with broken rotor bars and the bearing defect.