A Rolling Bearing Fault Diagnosis Method Based on Improved Cheetah Optimization Algorithm

Aiming at the problems of difficult extraction of nonlinear signal features and low diagnostic accuracy during rolling bearing faults as well as the slow convergence speed and insufficient optimization ability of the cheetah optimization algorithm, this paper proposes a rolling bearing fault diagnosis method (ICO-VMD-SE-DBN) based on the improved cheetah optimization algorithm and variational modal decomposition of deep confidence network. First, a mixed strategy improvement is made for the standard cheetah optimization algorithm, and the superiority of ICO is verified in the public test function. At the same time, using the minimum sample entropy (SE) as the objective function, the ICO algorithm is used to automatically find the optimization of the important parameters of the VMD, and the sample entropy is extracted from each component, which is used as the key eigenvalue of the signal, and second, the ICO algorithm is optimized to the hyper-parameters of the deep confidence network (DBN) to obtain the optimized ICO-DBN model. The final results show that the average accuracy of the method is as high as 99.86% in the public dataset, so the method can not only diagnose the rolling bearing fault state more accurately, but also has the same advantage in generalization and anti-interference.