Rolling element bearings are frequently used in rotary machinery, but they are also fragile mechanical parts. Hence, exact condition monitoring and fault diagnosis for them plays an important role of ensuring machineryâ€™s reliable running. Timely diagnosis of early bearing faults is desirable, but the early fault signatures are easily submerged in noise. This paper focuses on the application of dictionary learning and sparse representation methods on bearing fault feature extraction and fault diagnosis. Two dictionary learning methods: K-SVD and shift-invariant dictionary learning (SIDL) are studied. The K-SVD algorithm takes its name from singular value decomposition (SVD), which is used for one atom update and repeated for K times during the dictionary learning stage. K-SVD is one appealing method because of its efficiency. SIDL reconstructs signals using basis atoms in all possible time shifts. SIDL is very suitable to extract periodic impulses. Simulation and experimental bearing signals are used for demonstration and comparison of fault feature extraction based on KSVD and SIDL.
Author(s) by submitting the manuscript to the International Journal of COMADEM agree to transfer the rights to COMADEM International UK with some exceptions as described in COPYRIGHT TRANSFER AGREEMENT available here.