Fault Diagnosis for the Planetary Gearbox Based on a Hybrid Dimension Reduction Algorithm

Abstract

A hybrid dimension reduction algorithm based on feature selection and kernel principal component analysis (KPCA) is proposed in this paper to better realize the classification of the planetary gearbox faults. Firstly, in order to reduce the redundancy of some unnecessary features in the sample to a greater extent and the complexity of the kernel matrix calculation, a multi-criterion feature selection method is used to eliminate the irrelevant features. Secondly, through KPCA, the nonlinear principal component of the selected features is built. Then, fault is recognized by put the feature subset into the SVM classification. The proposed algorithm is applied to a planetary gearbox fault diagnosis experiment, and the experimental results show that the proposed algorithm outperforms the ones which employ feature selection or KPCA separately.