Multi-Layered Graph Convolutional Network-Based Industrial Fault Diagnosis with Multiple Relation Characterization Capability

Fault diagnosis of industrial equipments is extremely important for the safety requirements modern production processes. Lately, deep learning (DL) has been mainstream fault tool due to its powerful representational ability in and flexibility. However, most existing DL-based methods may suffer from two drawbacks: Firstly, only one metric used construct networks, thus multiple kinds potential relationships between nodes are not explored. Secondly, there few studies on how obtain better node embedding by aggregating features different neighbors. To compensate these deficiencies, an advantageous intelligent scheme termed AE-MSGCN proposed, which employs graph convolutional networks (GCNs) multi-layer innovative manner. In detail, AE carried out extract representation process measurement then combined with metrics (i.e., K-nearest neighbors, cosine similarity, path graph) interaction characterization among nodes. After that, intra-layer inter-layer adopted extensive neighbouring information enrich performance. Finally, a benchmark platform real-world case both verify that proposed more effective practical than state-of-the-art methods.