A Fault Diagnosis Method for Rotating Machinery Based on Gramian Angular Field and DGCN

The traditional fault feature extraction methods have large computational costs and are difficult to comprehensively characterize the signal state. To improve the fault diagnosis accuracy of rotating mechanisms such as rolling bearings, a fault diagnosis method based on Gramian Angular Field (GAF) and Dual-Channel Graph Convolutional Neural Network (DGCN) is proposed. Firstly, the signal is decomposed by GWO-VMD and the signal noise is filtered out by the kurtosis criterion. Then, the GAF algorithm is introduced to convert the one-dimensional time-domain signal into a two-dimensional image, thereby transforming the signal pattern diagnosis problem into an image classification problem. On this basis, combined with the characteristics of DGCN, this model is introduced into the classification of two-dimensional images to accurately identify faults. The Case Western Reserve University bearing dataset and the University of Connecticut gear dataset are introduced for case analysis. The results show that the proposed model can achieve an accuracy rate of 99.8%. Compared with different feature inputs and diagnostic models, the proposed method has a higher diagnostic accuracy rate and faster calculation speed, demonstrating certain superiority.