Rolling bearing fault diagnosis method based on RSSD and improved multi-population quantum genetic algorithm

In the process of resonance sparse signal decomposition, the choice of quality factor has an important influence on the decomposition result. At present, the Resonance Sparse Signal Decomposition (RSSD) mainly relies on the global optimization performance of artificial and genetic algorithms to select quality factors. However, conventional genetic algorithms are prone to fall into local optimal and premature convergence. In order to solve the problems of conventional genetic algorithms, in this paper, an improved multi-population quantum genetic algorithm is proposed to optimize the quality factor in resonance sparse signal decomposition. Compared with the existing methods, this method uses dynamic quantum revolving gates and catastrophic migration functions to adaptively obtain quality factors that match the characteristics of the original signal. Applying our proposed method to the early fault diagnosis of rolling bearings, the frequency amplitude of the fault signal obtained is about 50% higher than that of the traditional algorithm, and the resolution is improved, which can accurately and effectively extract the fault characteristic information in the vibration signal.