有监督不相关正交局部保持映射故障辨识  

Fault identification method based on supervised uncorrelated orthogonal locality preserving projection

作　　者：李锋[1] 王家序[2] 杨荣松[1] 

机构地区：[1]四川大学制造科学与工程学院,成都610065 [2]四川大学空天科学与工程学院,成都610065

出　　处：《仪器仪表学报》2013年第5期1113-1120,共8页Chinese Journal of Scientific Instrument

基　　金：四川大学青年教师科研启动基金(2012SCU11051);国家自然科学基金重点项目(50735008)资助

摘　　要：提出基于有监督不相关正交局部保持映射(SUOLPP)维数化简的故障辨识方法。首先构造全面表征不同故障特征的时频域特征集,再利用SUOLLP将高维时频域特征集自动约简为具有更好区分度的低维特征矢量,并输入到Littlewood-Paley小波核支持向量机中进行故障模式辨识。时频域特征融集可较全面准确地反映旋转机械的故障特征;SUOLPP同时利用流形局部几何结构和类标签来设计相似加权矩阵,并使输出基向量统计不相关和相互正交,提高了故障辨识精度。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 diagnosis method based on dimensionality reduction with supervised uncorrelated orthogonal Locality Preserving Projection （SUOLPP） is proposed in this paper. The time-frequency domain feature set is firstly constructed, which completely characterize the properties of different faults. Then, SUOLPP is used to automatically reduce the high-dimensional time-frequency domain feature sets of training and test samples to the low-dimensional eigenveetors that have better discrimination. Finally, the low-dimensional eigenvectors of training and test samples are inputted into Littlewood-Paley wavelet support vector machine （LPWSVM） to carry out fault identification. The time- frequency domain feature set can more comprehensively and accurately reflect the fault features of rotating machinery. SUOLPP uses both manifold local geometric structure and class labels to design the similarity weight ma- trix, and makes the output basis vectors statistically uncorrelated and orthogonal;therefore, higher fault identification accuracy is achieved. Fault diagnosis examples of deep groove ball bearing and life state identification of one type of space bearing demonstrate the effectivity of the proposed fault identification method.

关 键 词：时频域特征集 有监督不相关正交局部保持映射 维数化简 流形学习 故障辨识 

分 类 号：TH165.3[机械工程—机械制造及自动化] TN911.2[电子电信—通信与信息系统]