Research on fault feature extraction of rotating machinery based on adaptive VMD and optimized CYCBD

• Published in: 2022 34th Chinese Control and Decision Conference (CCDC) pp. 552 - 557
• Main Authors: Xie, Ran, Xiong, Ling, Dan, Binbin, Ren, Zeyu
• Format: Conference Proceeding
• Language: English
• Published: IEEE 15.08.2022
• Subjects: Electric shock; Fault diagnosis; Fault feature extraction; Feature extraction; Interference; Linear programming; Maximum second-order cyclostationarity blind deconvolution; Noise reduction; Teager energy operator; Variational mode decomposition; Vibrations; Wavelet threshold denoising
• DOI: 10.1109/CCDC55256.2022.10034111

To solve the problem that early fault features of rotating machinery are difficult to extract, an adaptive k-value hierarchical variational mode decomposition (H-VMD) combined with optimized maximum second-order cyclostationarity blind deconvolution (CYCBD) fault feature extraction method is proposed in this paper. Mode decomposition of the vibration signal is performed with H-VMD. and then the noise dominant component is denoised by wavelet threshold denoising (WTD). Furthermore, the improved autocorrelation-weighted correlated kurtosis (ACK) when CYCBD enhances the periodic shock component of the denoised signal is the fitness function of ChOA, and the envelope demodulation analysis of the feature-enhanced signal is performed using the teager energy operator (TEO). Simulation analysis and experimental results show that the interference of background noise can be effectively removed and the periodic shock component of the vibration signal be enhanced by the proposed method, which is a new feature extraction method for the fault diagnosis of rotating machinery.