Bearing Fault Diagnosis Method Based on Extra-perceptual Graph Neural Network
CHEN Yanyan
ZHU Yanmin
Abstract：The traditional graph neural network did not consider the problems of difficult feature extraction and weak extraction ability under the condition of fault,noise and vibration waveform changes,which led to low bearing fault diagnosis accuracy under real working conditions.In order to solve this problem,a bearing fault diagnosis method based on extra-perceptual graph neural network(E-GCN)was proposed.In this method,the dilation residual module was proposed,which could effectively expand the receptive field and learn richer and more effective features,thereby improving its performance and generalization ability.The method constructed additional figure convolution module to obtain the global information of the graph,and it also used the attention module to weight the features of the neighbor nodes,so that it could focus more on the important features,which further improved the accuracy and efficiency of diagnosis.The experiments were carried out on the bearing dataset of Paderborn University(PU)and Case Western Reserve University(CWRU),and the diagnostic accuracy rate reached 98.2％and 99.1％,respectively.This method can be effectively applied to the field of fault diagnosis of bearings and other machinery.
Keywords：fault diagnosisextra-perceptualdilation residualsattention mechanismsgraph neural networkbearing
Publication Date：2024-12-20
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：7( 500-506 )