基于低秩稀疏分解算法的航空锥齿轮故障诊断

锥齿轮是航空发动机传动系统改变传动方向和传递功率的核心部件,常工作于高速、重载的条件下,不可避免发生齿面损伤及疲劳断裂等故障,其故障模式复杂、背景噪声强烈,难以有效诊断。针对上述问题,研究了旋转机械特征的自相似属性,分析了观测信号中干扰信息和特征信息的奇异值分布差异性,建立了二维特征矩阵的奇异值稀疏低秩先验,并将其建模为稀疏核范数正则描述,进而构建了低秩稀疏分解模型,提出了基于广义块坐标优化理论的模型求解算法框架。将算法用于滑油附件锥齿轮的故障诊断,有效地辨识了潜在的故障,并与经典的稀疏正则和标准的谱峭度算法进行对比分析,结果证实了算法的优越性。

Fault diagnosis of aero-engine bevel gear based on a low rank sparse model

Bevel Gear is a key component in aero-engine transmission system, which always works in harsh environment such as high speed and high load. Therefore, they inevitably suffer performance degradation. However, the observed signals are also contaminated by strong background noises and harmonic interferences. In the paper，a novel low rank sparse decomposition method is proposed for aero-engine bevel gear fault diagnosis. Firstly, due to the self-similarity of impulsive feature, an adaptive partition window is designed to transform the impulsive feature into a data matrix. By performing the SVD decomposition, the singular value distribution of feature signal exhibits sparse property, and then the sparse low rank prior of feature signal is established, which is further modeled by nuclear norm. Subsequently, by incorporating the classic sparse learning model and the nuclear norm of feature signal, a novel sparse low rank model is proposed. Furthermore, a proximal gradient basedonblock coordinate decent solver is also developed. The effectiveness of the proposed model and algorithm are evaluated through performing the diagnosis of aero-engine bevel gear.