Research on fault diagnosis of rotating machinery based on deep learning and SMOTE data enhancement
LI Wenzan
LIU Qinming
YE Chunming
WANG Yujie
Abstract：[Objective]With the advancement of automation and intelligence in industrial equipment,fault diagnosis of rotating machinery has become a critical component in ensuring the stable operation of equipment.Traditional diagnostic approaches,which often depend on expert knowledge and basic signal processing techniques,struggle to manage complex operating conditions and varied fault types.To address these challenges,a novel fault diagnosis framework was introduced,leveraging Gram angular field(GAF)image encoding and an enhanced gated WaveNet(GWaveNet)architecture.[Methods]Firstly,the original one-dimensional bearing vibration signals were transformed into two-dimensional images via GAF encoding,effectively preserving temporal dependencies and dynamic signal variations.In this process,to mitigate issues related to data imbalance,techniques such as window segmentation,noise augmentation,and synthetic minority over-sampling technique(SMOTE)were employed,thereby increasing sample diversity and improving model's robustness and training efficacy.Secondly,the enhanced GWaveNet architecture integrated convolutional layers,residual connections,and multi-scale feature extraction mechanism,which collectively strengthen the network's capacity to recognize different fault patterns.Finally,test validation on the Case Western Reserve University(CWRU)public dataset was performed.The test results demonstrated that the proposed method can achieve high diagnostic accuracy under three operating conditions—normal,inner race fault,and outer race fault,while exhibiting strong generalization performance in complex operating conditions.[Results]The results demonstrate that the proposed model exhibits high accuracy and robustness in fault diagnosis of rotating machinery,and can effectively address a wide range of fault types and uncertainties inherent in real industrial environments.
Keywords：Rotating machineryFault diagnosisDeep learningGram angular fieldData enhancement
Publication Date：2026-02-15
Online Publishing Date：2026-09-12(First online date of this platform, not the publication date of the document)
Pages：11( 1-11 )