Multi-fault diagnosis for rolling element bearings based on intrinsic mode function screening and optimized least squares support vector machine

• Main Authors: Qingbin Tong, Baozhu Han, Yuyi Lin, Weidong Zhang
• Format: Article
• Language: English
• Published: JVE International 2016-11-01
• Series: Journal of Vibroengineering
• Subjects: fault diagnosis; redundant second generation wavelet packet transform (RSGWPT); ensemble empirical mode decomposition (EEMD); intrinsic mode functions (IMFs); energy ratio; singular value decomposition (SVD); least squares support vector machine (LSSVM); particle swarm optimization (PSO)
• Online Access: https://www.jvejournals.com/article/17090
• ISSN: 1392-8716; 2538-8460

Multi-fault diagnosis of rolling element bearing is significant to avoid serious accidents and huge economic losses effectively. However, due to the vibration signal with the character of nonstationarity and nonlinearity, the detection, extraction and classification of the fault feature turn into a challenging task. This paper presents a novel method based on redundant second generation wavelet packet transform (RSGWPT), ensemble empirical mode decomposition (EEMD) and optimized least squares support vector machine (LSSVM) for fault diagnosis of rolling element bearings. Firstly, this method implements an analysis combining RSGWPT-EEMD to extract the crucial characteristics from the measured signal to identify the running state of rolling element bearings, the vibration signal is adaptively decomposed into a number of modified intrinsic mode functions (modified IMFs) by two step screening processes based on the energy ratio; secondly, the matrix is formed by different level modified IMFs and singular value decomposition (SVD) is used to decompose the matrix to obtain singular value as eigenvector; finally, singular values are input to LSSVM optimized by particle swarm optimization (PSO) in the feature space to specify the fault type. The effectiveness of the proposed multi-fault diagnosis technique is demonstrated by applying it to both simulated signals and practical bearing vibration signals under different conditions. The results show that the proposed method is effective for the condition monitoring and fault diagnosis of rolling element bearings.