Combining Support Vector Machine – Fast Fourier Transform (SVM – FFT) For Improving Accuracy on Broken Bearing Diagnosis

Electric motor has critical component that called as bearing. Bearing condition be monitored through vibration signal that is produced by vibration sensor. Vibration signal is analysed to detect condition of bearing and also to diagnose the broken bearing. In order to intelligently diagnose the broken bearing, signal generated by vibration -usually mentioned as vibration signal is utilized. Currently, machine learning is spready utilized to diagnose the broken bearing either by utilizing shallow architecture or deep architecture. This report presents the result of improving accuracy on diagnosing Broken Bearing using combined Support Vector Machines (SVM) and Fast Fourier Transform (FFT). The result showed that by using SVM - FFT could outperform combined SVM - Statistic and also outperform others shallow classifiers such as Decision Tree, Random Forest, and Naive Bayes. The combined SVM - FFT could be used for diagnosing Broken Bearing with higher accuracy (accuracy> 99%).