Evaluating the Progression and Orientation of Scratches on Outer-Raceway Bearing Using a Pattern Recognition Method

Bearing faults are a major source of failure in an induction motor, and early detection of fault becomes necessary because of its industrial application. A range of analytical methods has been used to detect, identify, and diagnose bearing faults, including vibrational analysis. Most analyses have used pitting as the fault, whereas in the industrial environment, scratches are a more common problem. This paper investigates such scratches, applying two types of fault analysis: fault progression and fault orientation. A support vector machine (SVM) algorithm is used to classify and diagnose the different types of bearing fault. The frequency-domain features obtained from a fast Fourier transform of the load current is used to train the SVM algorithm. The proposed diagnostic method is tested experimentally using induced outer race faults under different load conditions. The method is shown to be successful in diagnosing faults, suggesting potential applications in real industrial settings.