Jixie chuandong (Jan 2023)
Application of Enhanced EWT and Enhanced Dictionary Learning in Bearing Faults Identification
Abstract
When realizing rolling bearing fault identification through deep learning, there is a low recognition rate and convergence rate due to ambient noise. Aiming at the above problem, a fault identification model based on enhanced empirical wavelet transform (EEWT) and enhanced dictionary learning (EDL) is proposed. Firstly, the vibration signals of rolling bearing are transformed by envelope spectrum, and envelope spectrum adaptive segmentation is implemented through the relationship between the envelope point and the adaptive threshold, and signals are decomposed into several amplitude modulation-frequency modulation (AM-FM) components. Secondly, a new component screening index is proposed, and then the appropriate AM-FM components are reconstructed to effectively reduce the noise of signals. Finally, the sparsity constraint is used to learn the typical structural characteristics in the bearing fault sample layer by layer, and the deep fault dictionary (DFD) is constructed. Then the fault samples are fed into the DFD to determine the fault category according to the reconstruction error of the samples. The test results show that the proposed method is robust to noise and has better fault recognition ability than other models. And the proposed method utilizes the sparse constraint driving dictionary to automatically extract the fault features in the vibration signal samples, while the EDL structure makes the extracted fault features have better hierarchical and physical meaning, which is in line with people's intuitive understanding of the fault and can be used in the rolling bearing fault identification engineering.
