Deep Adaptively Dynamic Edge Graph Convolution Network with Attention Weight and High-dimension Affinity Feature Graph for Rotating Machinery Fault Diagnosis

Abstract Graph neural network (GNN) has emerged as an effective way to mine relationship between data due to its powerful modeling capability for structure data, and have garnered significant attention from scholars for intelligent fault diagnosis tasks. However, the adjacency matrix of mostly GNN models with deep architecture are always fixed during aggregation process, and the edge connection relationship cannot be adaptively adjusted, which limits their performance for feature representation. Besides, for few-shot diagnosis scenarios of rotating machinery, the generalization performance of deep GNN models will be further degraded due to fixed receptive fields and limited training samples. To address these issues, a deep adaptively dynamic edge graph convolution network (DADE-GCN) with attention weight and high-dimension affinity feature graph is proposed. First, a deep adaptively dynamic edge graph convolution module with attention weight is developed to dynamically adjust the receptive field in different graph convolution layers by adaptively changing the nearest neighbors and edge connection relationship to construct new adjacency matric. Subsequently, the output features of different layers are fused by self-attention mechanism. Second, to overcome the effect of time-shift problem existing in vibration signals and capture accurate interdependencies between data under few-shot diagnosis tasks, a high-dimension affinity feature graph construction method is proposed to construct graph structure data. The effectiveness of proposed method is quantitatively verified by two rotating machinery datasets, indicating that the proposed DADE-GCN model can achieve the average diagnosis accuracies of 99.83% and 98.80% in the few-shot diagnosis tasks, which is significantly superior than several state-of-the-art recognition methods.