Condition Monitoring and Diagnosis of Outer Raceway Bearing Fault using Support Vector Machine

Fault detection at initial stage and follow-up the monitoring progress are mandatory to avoid the damage of induction motors. If fault not detected at initial stage, it may lead to economic loss or even shutdown of the industry. Some of the commonly occurring bearing faults are scratch, fluting and pitting. From industrial point of view, the possibility of multiple fault must be considered, as a progressive stage of initial single failure. Thus, in the present study, three types of bearing conditions (hole, scratch, hole and scratch) are discussed. First, fast Fourier transform (FFT) is performed using the load current of stator. Next, features of frequency spectrum are extracted and finally diagnosis is carried out using Support Vector Machine (SVM) algorithm. The algorithm is trained using the features of FFT analysis. Finally, accuracy of the diagnosed results is discussed.