Compound Fault Diagnosis of Rolling Bearing Based on Tunable Q-Factor Wavelet Transform and Sparse Representation Classification

Rolling bearing is a vital component of many mechanical equipment, but it’s easy to damage, especially in harsh working conditions for a long time. This damage is either caused by a single failure or caused by a composite failure. The paper presents a new compound fault diagnosis method of rolling bearing, in which Wavelet Transform (WT) for feature extraction and Sparse Representation based Classification (SRC) for diagnosis are comprehensively applied. Firstly, the Tunable Q-Factor WT is performed on bearing vibration signals to extract fault features at different frequency bands. Then, the fault features in each band are coded sparsely on the established training sample dictionary set, and the respective fault feature bands are reconstructed using the sparse coefficient. Finally, the composite fault type of the bearing fault is judged according to the fault category where the minimum value is located. The effectiveness of the proposed method is verified by the simulation experiment and the bearing failure experiment.