A novel fault diagnosis method based on improved adaptive variational mode decomposition, energy entropy, and probabilistic neural network

Abstract

To improve the accuracy of bearing fault recognition, a novel bearing fault diagnosis (PAVMD-EE-PNN) method based on parametric adaptive variational mode decomposition (VMD), energy entropy, and probabilistic neural network (PNN) is proposed in this paper. In view of the effect of VMD on signal decomposition effect affected by the number of preset decomposition modes, a central frequency screening method is proposed to determine the number of decomposition modes of the VMD method. The parametric adaptive VMD method is used to decompose the bearing fault signal into a series of intrinsic mode function (IMF) components. The energy entropy of IMF components is calculated to form an eigenvector, which is input into the PNN model for training to obtain a fault recognition model with maximum output probability. The actual bearing vibration data are obtained and used to test and verify the effectiveness of the PAVMD-EE-PNN method. The experimental results show that the PAVMD-EE-PNN method can effectively and accurately identify the fault type, and the fault recognition effect is better than contrast fault diagnosis methods.