Rotating Machinery Fault Diagnosis Based on Manifold Learning using Semi-supervised Local Linear Embedding

It is convincingly demonstrated that manifold learning can be employed to extract fault feature, and can be combined with pattern classifier to accomplish fault diagnosis. In this paper, a novel fault diagnosis approach based on semi-supervised local linear embedding (SSLLE) which extracts the low-dimensional manifold embedded in high-dimension fault data by taking advantage of the class label information is proposed. This fault diagnosis approach firstly extracts the intrinsic manifold features using SSLLE, and has a higher accuracy and better performance compared to traditional algorithms. Experiments on the fault diagnosis for rotating machinery certify the practicability and effectiveness of the proposed algorithm.