TD-GAT: Graph Neural Network for Fault Diagnosis Knowledge Graph
Yang Liu, Yajie Ma, Zehui Mao, Bin Jiang, Jianwei Liu, Jianan Bian · 2021 China Automation Congress (CAC) · 2021
As an important resource for artificial intelligence applications, the ultimate goal of the knowledge graph (KG) is to make artificial intelligence smarter. In recent years, graph representation learning and graph neural networks are essential to the improvement and application of knowledge graph. In this paper, in order to solve the fault diagnosis problem of the system at runtime and realize the intelligent auxiliary decision-making of fault diagnosis, we introduce the knowledge graph technology. However, according to existing research, although some papers have used knowledge graphs in the field of fault diagnosis, they are only limited to use existing fault knowledge data to build knowledge graphs. therefore, this paper proposes a new method of knowledge graph embedding and knowledge reasoning for fault diagnosis knowledge graph, called TransD Graph Attention Network (TD-GAT). This model (TD-GAT) explicitly learns the entity representation of the knowledge graph in an end-to-end learning mode. Finally, this paper uses the fault entity classification task as an example to demonstrate the superiority of our model (TD-GAT) in the fault diagnosis knowledge graph we have established.