ROLLER BEARING FAULT DETECTION USING EMPIRICAL MODE DECOMPOSITION AND ARTIFICIAL NEURAL NETWORK METHODS

One of the methods for detection faults in structural and mechanical systems is processing vibrational signals extracted from the real system. The Hilbert-Huang transform (HHT) is a new and strong method for analyzing nonlinear and non-stationary vibrations based on time-frequency This approach is based on decomposing a signal into empirical modes and Hilbert spectral analysis. In the current paper first, vibrational signals of a roller bearing are decomposed into intrinsic mode functions (IMFs) using ensemble empirical mode decomposition (EEMD) method and IMFs sensitive to impulse are determined by Kurtosis coefficient. Then Kurtosis and standard deviation factors are extracted from the mentioned IMFs and used for training and validating the multi layers perceptron neural network. The results of network trial showed faulty or normal roller hearing and its fault type. (C) 2019 INT TRANS J ENG MANAG SCI TECH.