Machine condition diagnosis is studied by combining the principal component analysis (PCA) and the subspace methods. In the principal component subspace-based method, two subspace structures, called information compression subspace and class-specific subspace, are presented to represent and classify machine condition pattern. The proposed methods can extract effective and stable low-dimensional features of machine condition by the PCA, and effectively identify machine condition with low computational complexity based on the subspace methods. Experimental analysis with fatigue condition diagnosis of an automobile transmission gearbox shows that two principal component subspace-based methods have achieved excellent results with respective merit, and can be effectively applied to machine condition monitoring and diagnosis.
HE Qing-bo; LIU Yong-bin; WU Qiang; LI Peng; HU Fei;KONG Fan-rang.
Machine condition diagnosis based on principal component subspace[J]. Journal of Vibration and Shock, 2011, 30(10): 265-269
PDF(2000 KB)
