A Method for Rolling Bearing Fault Diagnosis Based on Sensitive Feature Selection and Nonlinear Feature Fusion

To solve the problem that there were non-sensitive features and over-high dimensions in the feature set of fault diagnosis, a new feature extraction method based on sensitive feature selection and nonlinear feature fusion for rolling element bearing fault diagnosis was proposed. CDET was utilized to choose features sensitive to fault severity from the high dimensional feature set, and weighted the selected sensitive features by their sensitive degree. The weighted sensitive feature subset was compressed with LPP to reduce its dimensions and get the compressed more sensitive feature subset, which can enhance the discrimination between all classes by improving the between-class scatter and within-class scatter for fault classification. Compared with the classification result of the original high dimensional feature set based on direct LPP without feature selection, the classification result of the proposed method has the advantages of higher compactness and higher computational efficiency.