Intrinsic Mode Function Feature Energy and PCA-based Fault Recognition Method of Rolling Bearing

摘要

Scientists working with fault diagnosis regularly confront the problem: vibrations from faulty parts are nonlinear and non-stationary. Here an approach was proposed to solve the problem by combination of Intrinsic Mode Function (IMF) energy and Principal Component Analysis (PCA). The approach firstly implements empirical mode decomposition to the vibration signals, from which IMFs are obtained. PCA reduces the dimensionality of feature vectors composed of IMF energy values, by which bearing conditions can be visually diagnosed. And BP neural network is used to classify fault types. The approach was confirmed by case study on bearing defect recognition. The results demonstrate that the approach can recognize fault types with higher accuracy in shorter time.