Bearing fault diagnosis based on EMD-KPCA and ELM

摘要

In recent years, many studies have been conducted in bearing fault diagnosis, which has attracted increasing attention due to its nonlinear and non-stationary characteristics.To solve this problem, this paper proposes, a fault diagnosis method based on Empirical Mode Decomposition (EMD), Kernel Principal Component Analysis (KPCA), and Extreme Learning Machines (ELM) neural network, which combines the existing self-adaptive time-frequency signal processing with the advantages of non-linear multivariate dimensionality reduction KPCA approach and ELM neural network.First, EMD is applied to decompose the vibration signals into a finite number of intrinsic mode functions, in which the corresponding energy values are selected as the initial feature vector.Second, KPCA is used to further reduce the dimensionality for a simplified low-dimension feature vector.Finally, ELM is introduced to classify the extracted fault feature vectors for lessening the human intervention and reducing the fault diagnosis time.Experimental results demonstrate that the proposed diagnostic can effectively identify and classify typical bearing faults.