Bearing Fault Diagnosis Using Attractor Comparison

The failure in rolling element bearings as an important part of rotary machine can lead to machine breakdown. Finding a good strategy for fault diagnosis can help to reduce maintenance cost and enhance reliability. The S-statistic based on the comparison of two reconstructed attractors (RA's) in the phase space was applied to diagnose different types of faults in bearings (inner, ball, and outer faults). To achieve the aim, measured signals of a normal bearing and several faulty bearings were decomposed to sub-signals by wavelet packet so that frequency of each defect is located in a unique packet. Comparing the result for normal and faulty sub-signals using the S-statistic indicate that the method can be applied for fault diagnosis.