부산대학교 도서관

The most common motor fault among electrical motors is a bearing fault. Different bearing faults produce different vibrations which can be recognized by machine learning algorithms. A real-time mechanical motor bearing fault diagnosis system using a Support Vector Machine classifier and Hilbert-Huang Transform as feature extraction was proposed in this study. The real-time motor bearing fault diagnosis model system was run using MATLAB software to communicate with an accelerometer as a vibration sensor. Six (6) bearing fault types were used to simulate different bearing faults. The average accuracy obtained during the testing of the model was 78.96%.