Research On Fault Diagnosis Of Motor Bearing Based On VMD And PNN

As an important driving equipment for industrial production,the safe and reliable operation of motor is a prerequisite to ensure production.Rolling bearing is one of the most important and vulnerable components in the motor,and its health status is very important for the normal operation of the motor,so it is important to carry out fault diagnosis research on the motor bearing.In this study,the fault signal of motor bearing is taken as the research object,and the signal processing and fault identification in bearing fault diagnosis are mainly studied as follows.Firstly,the basic structure and common failure forms of bearings as well as the current development status of fault diagnosis are explained in this paper.The basic principle of Variational Mode Decomposition(VMD)is studied in depth.In order to show the effectiveness of the VMD method in processing signals,a comparative analysis with the Empirical Mode Decomposition(EMD)method is conducted to show the superiority of the VMD method through the decomposition results obtained by both methods.The influence of the important parameters in the VMD method on its decomposition results is also studied.Secondly,the influence of the number of decompositions and penalty factors on the decomposition results is investigated for the parameter selection problem of the VMD method in the process of signal processing,and a VMD parameter optimization method based on the improved Beetle Antennae Search(BAS)algorithm is proposed.A suitable combination of parameters is determined according to the characteristics of the signal itself,and the decomposition effect of the VMD method is better than that of the EMD method under this combination of parameters.In order to improve the accuracy of bearing fault classification,a fault identification method based on sensitive features and improved BAS optimized Probabilistic Neural Network(PNN)is proposed.The superior performance of the Locally Linear Embedding(LLE)algorithm in signal data dimensionality reduction is obtained through comparative experiments,and the sensitive feature parameters in the reconstructed signal are obtained to characterize the operating status of the bearing using the LLE algorithm.Since the accuracy of PNN network classification depends on the smoothing coefficient,the parameters are selected using the improved BAS algorithm,and a fault identification method based on sensitive features and improved BAS optimized PNN is proposed in combination with the above data dimensionality reduction method.Finally,the signals of four operating states of laboratory motor bearings are used as the research object,and the VMD of various types of signals optimized by the improved BAS is decomposed,and the effective IMF components are reconstructed according to the cliffness,and the sensitive features in the time and frequency domain statistical features of the reconstructed signals are extracted by combining LLE to complete the training and testing of the PNN network optimized by the improved BAS algorithm.The experimental results verify the feasibility and high fault recognition rate of the motor bearing fault diagnosis method proposed in this paper,which has certain research significance.