Bearing and Gear Fault Diagnosis Using Adaptive Wavelet Transform of Vibration Signals

The continuous wavelet transform (CWT) is widely used for analyzing non-stationary vibration signals from rotating machines. It is a well known fact that, if the shape of the wavelet function matches adequately with the shape of the burst present in signal at a specific scale and location, a large transform value is obtained, and a low transform value is obtained if the signal and wavelet do not correlate well. It is also worth noticing that each rotating system and its corresponding defect signature are unique in nature. Therefore, the challenge remains in arriving with a robust wavelet function for the wavelet transform technique to solve any specific class of problem in hand. This challenge has been addressed in the present work and in order to understand the defect signature appropriately, the concept of adaptive wavelet design and its implementation for defect identification are discussed. In presence of seeded defects, a random burst from the vibration signal is used for designing the corresponding adaptive wavelet. Similarly, the designed adaptive wavelet is used to compute the CWT coefficients. The CWT coefficients so generated are compared with the standard wavelet based scalograms. The adaptive wavelet transform is shown to be apposite in analyzing the vibration signals and also is corroborated with the acquired experimental data.