Event State Knowledge Graph-driven Full-chain Product Quality Tracing

Product quality tracing is a core function of manufacturing execution systems. It involves tracking and feedback control of product quality within an enterprise’s supply chain. This paper proposes a full-chain product quality tracing method using an event state knowledge graph. The method uses the Input-Output data of the nodes in the supply chain as the data source for the knowledge graph. It achieves the correlation of the supply chain’s multivariate heterogeneous data and then constructs the knowledge graph according to a top-down approach. The knowledge graph’s schema layer is first built based on ontology and semantic networks during construction. Then, the data layer of the knowledge graph for product quality tracing is automatically generated by introducing an event-state dynamic triggering mechanism. Next, the graph neural network calculates the edge weights of the knowledge graph and the weights of neighbouring nodes to the central node. This enables point tracing, chain tracing, network tracing, and feedback control of product quality. Finally, the method’s feasibility is verified through a case study of an electronics industry supply chain in a certain high-tech development zone.