A fault diagnosis of bearing through SVM classifier and data visualization using PCA

In modern industrial circumstances, there is a huge demand for condition based monitoring for the induction motor. Amongst all the elements of the induction motor, bearing is the most critical element and it‟s prone to fail which causes a major issue. Usually, vibration analysis is used to detect the fault in the bearing. However, this analysis often leads to difficulty to access the locations of the equipment and also it is costly. In this paper, fault detection in the bearing of three-phase induction motor was performed by extracting the time domain features, wavelet energy features and wavelet entropy features. For the diagnosis of bearing failures, the extraction of features must be performed to extract numerous critical features from the high-dimensional data acquired from the vibration signals by performing a time-domain and time-frequency domain that would convert complicated signals into prominent low-dimensional features through PCA algorithms. Classification was evaluated by the Support Vector Machine. Results signify that the suggested approach ameliorate classification efficiency using wavelet energy and wavelet entropy features than the time domain features.