Fault detection and diagnosis on the rolling element bearing

The use of novel sensors such as Air-coupled ultrasonic transducer, Eddy current, Piezoelecteric ultrasonic transducer, and piezo strain gage as diagnostic tools for detection of bearing faults have been investigated. A series of experiments was carried out in a laboratory environment. Localized defects with different sizes were created intentionally on the test bearing components simulating evolving cracks or other related faults. Four different signal processing techniques were applied to extract the signal features. The resulting data for different bearing speed and load showed that the sensors are capable of detecting different types of defects located on the bearing components. The data from the sensors were also compared with those obtained from accelerometers. Then, a practical classification model based on a decision tree algorithm was developed and applied on the extracted features. The results showed that the defect could be detected and classified at an early stage on each component of the bearing.