Hybrid fault-feature extraction of rolling element bearing via customized-lifting multi-wavelet packet transform

The rolling element bearing is one of the most extensively used components in various rotating machinery, and it is therefore critical to develop a suitable online rolling element bearing fault-diagnostic framework to improve a rolling element bearing system's failure protection during conditional operations. In this paper, a hybrid fault-feature extraction method by detecting localized defects and analyzing vibration signals of rolling element bearings via customized multi-wavelet packet transform is proposed, in which the swarm fish algorithm has been utilized for the minimization of signal residual to determine the adaptive prediction operator. With good properties of concurrent symmetry, orthogonality, short support and high-order vanishing moment, the multiple wavelet functions and scaling functions are defined for the hybrid fault-feature extraction, which match the diverse characteristics of hybrid fault and extract coupling features, and the proposed lifting scheme-based multi-wavelet packet transform is highly effective. Then, the proposed method is validated by rolling element bearing experimental results, which show that this approach can effectively extract the hybrid fault features of inner race and rolling element.