组稀疏低秩矩阵估计的变转速滚动轴承故障特征提取

Effective extraction of early bearing failure features is of great importance to avoid serious mechanical accidents. The impulse signals characterizing bearing faults are often submerged in strong background noise interference, and the bearings often operate under variable speed conditions, which makes the task of fault feature extraction more difficult. To address this issue, one kind of group sparse low-rank matrix estimation algorithm for rolling bearing fault feature extraction under variable speed conditions was proposed in this paper. Firstly, the measured signal was transformed into the order-frequency domain by using order-frequency spectral correlation (OFSC) according to the angle/time cyclist-ationarity of the bearing fault pulse signal under variable speed conditions. Secondly, the group sparsity and low-rank property of the bearing fault pulse in the order-frequency domain were revealed, and a convex optimization problem was constructed to enhance these two properties accordingly, and a non convex penalty function was introduced to improve the sparsity of the fault characteristics. Again, the convex optimization problem was solved in the framework of the alternating direction method of multipliers (ADMM) and majorization-minimization (MM), and a group sparse low-rank (GSLR) matrix estimation algorithm was derived. Finally, the target components obtained from the solution were detected by constructing the enhanced envelope order spectrum (EEOS) for the fault features. The analysis of simulation and experimental signals verify the effectiveness of the method in fault feature extraction.