基于有监督增量式局部线性嵌入的故障辨识

Fault identification method based on supervised incremental locally linear embedding

提出一种基于有监督增量式局部线性嵌入的故障辨识方法。首先构造全面表征不同故障特性的时频域特征集,再利用有监督增量式局部线性嵌入将高维时频域特征集自动化简为区分度更好的低维特征矢量,并输入Morlet小波支持向量机中进行故障模式辨识。有监督增量式局部线性嵌入结合流形局部几何结构和类标签来设计重构权值矩阵,并采用局部线性投影计算新增样本的嵌入映射,提高了故障辨识精度,实现了新样本的快速增量处理。深沟球轴承故障诊断和空间轴承寿命状态辨识实例验证了该方法的有效性。

A novel fault identification method based on supervised incremental locally linear embedding (SILLE) was proposed here. The time-frequency domain feature set was firstly constructed to completely characterize the property of each fault. Then, SILLE was introduced to automatically simplify high-dimensional time-frequency domain feature sets of training and test samples into low-dimensional eigenvectors having better discrimination. Finally, the low-dimensional eigenvectors of training and test samples were input into Morlet wavelet support vector machine (MWSVM) to perform fault identification. With SILLE, both local manifold geometry and class labels were combined to design the reconstruction weighted matrix, local linear projections were adopted to obtain the embedded mapping of the new fault samples. Thus, the fault identification accuracy was improved and the rapid incremental processing of the new samples was realized. Fault diagnosis of deep groove ball bearings and life state identification of one type of space bearings demonstrated the effectiveness of the proposed fault identification method.